Transhumanismus

A new system for forecasting weather and predicting future climate uses artificial intelligence to achieve results comparable with the best existing models while using much less computer power, according to its creators.

In a paper published in Nature yesterday, a team of researchers from Google, MIT, Harvard, and the European Center for Medium-Range Weather Forecasts say their model offers enormous “computational savings” and can “enhance the large-scale physical simulations that are essential for understanding and predicting the Earth system.”

The NeuralGCM model is the latest in a steady stream of research models that use advances in machine learning to make weather and climate predictions faster and cheaper.

What Is NeuralGCM?

The NeuralGCM model aims to combine the best features of traditional models with a machine-learning approach.

At its core, NeuralGCM is what’s called a “general circulation model.” It contains a mathematical description of the physical state of Earth’s atmosphere and solves complicated equations to predict what will happen in the future.

However, NeuralGCM also uses machine learning—a process of searching out patterns and regularities in vast troves of data—for some less well-understood physical processes, such as cloud formation. The hybrid approach makes sure the output of the machine learning modules will be consistent with the laws of physics.

The resulting model can then be used for making forecasts of weather days and weeks in advance, as well as looking months and years ahead for climate predictions.

The researchers compared NeuralGCM against other models using a standardized set of forecasting tests called WeatherBench 2. For three- and five-day forecasts, NeuralGCM did about as well as other machine-learning weather models such as Pangu and GraphCast. For longer-range forecasts, over 10 and 15 days, NeuralGCM was about as accurate as the best existing traditional models.

NeuralGCM was also quite successful in forecasting less-common weather phenomena, such as tropical cyclones and atmospheric rivers.

Why Machine Learning?

Machine learning models are based on algorithms that learn patterns in the data fed to them and then use this learning to make predictions. Because climate and weather systems are highly complex, machine learning models require vast amounts of historical observations and satellite data for training.

The training process is very expensive and requires a lot of computer power. However, after a model is trained, using it to make predictions is fast and cheap. This is a large part of their appeal for weather forecasting.

The high cost of training and low cost of use is similar to other kinds of machine learning models. GPT-4, for example, reportedly took several months to train at a cost of more than $100 million, but can respond to a query in moments.

A comparison of how NeuralGCM compares with leading models (AMIP) and real data (ERA5) at capturing climate change between 1980 and 2020. Credit: Google Research

A weakness of machine learning models is that they often struggle in unfamiliar situations—or in this case, extreme or unprecedented weather conditions. To improve at this, a model needs to generalize, or extrapolate beyond the data it was trained on.

NeuralGCM appears to be better at this than other machine learning models because its physics-based core provides some grounding in reality. As Earth’s climate changes, unprecedented weather conditions will become more common, and we don’t know how well machine learning models will keep up.

Nobody is actually using machine learning-based weather models for day-to-day forecasting yet. However, it is a very active area of research—and one way or another, we can be confident that the forecasts of the future will involve machine learning.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Image Credit: Kochov et al. / Nature

Age catches up with us all. Eyes struggle to focus. Muscles wither away. Memory dwindles. The risk of high blood pressure, diabetes, and other age-related diseases skyrockets.

A myriad of anti-aging therapies are in the works, and a new one just joined the fray. In mice, blocking a protein that promotes inflammation in middle age increased metabolism, lowered muscle wasting and frailty, and reduced the chances of cancer.

Unlike most previous longevity studies that tracked the health of aging male mice, the study involved both sexes, and the therapy worked across the board.

Lovingly called “supermodel grannies” by the team, the elderly lady mice looked and behaved far younger than their age, with shiny coats of fur, less fatty tissue, and muscles rivaling those of much younger mice.

The treatment didn’t just boost healthy longevity, also known as healthspan—the number of years living without diseases—it also increased the mice’s lifespan by up 25 percent. The average life expectancy of people in the US is roughly 77.5 years. If the results translate from mice to people—and that’s a very big if—it could mean a bump to almost 97 years.

The protein, dubbed IL-11, has been in scientists’ crosshairs for decades. It promotes inflammation and causes lung and kidney scarring. It’s also been associated with various types of cancers and senescence. The likelihood of all these conditions increases as we age.

Among a slew of pro-aging proteins already discovered, IL-11 stands out as it could make a beeline for testing in humans. Blockers for IL-11 are already in the works for treating cancer and tissue scarring. Although clinical trials are still ongoing, early results show the drugs are relatively safe in humans.

“Previously proposed life-extending drugs and treatments have either had poor side-effect profiles, or don’t work in both sexes, or could extend life, but not healthy life, however this does not appear to be the case for IL-11,” said study author Dr. Stuart Cook in a press release. “These findings are very exciting.”

Strange Coincidence

In 2017, Cook zeroed in on IL-11 as a treatment target for heart and kidney scarring, not longevity. Injecting IL-11 triggered the conditions, eventually leading to organ failure. Genetically deleting the protein protected against the diseases.

It’s easy to call IL-11 a villain. But the protein is an essential part of the immune system. Produced by the bone marrow, it’s necessary for embryo implantation. It also helps certain types of blood cells grow and mature, notably those that stop bleeding after a scrape.

With age, however, the protein tends to goes rogue. It sparks inflammation across the body, damaging cells and tissues and contributing to cancer, autoimmune disorders, and tissue scarring. A “hallmark of aging,” inflammation has long been targeted as a way to reduce age-related diseases. Although IL-11 is a known trigger for inflammation, it hasn’t been directly linked to aging.

Until now. The story is one of chance.

“This project started back in 2017 when a collaborator of ours sent us some tissue samples for another project,” said study author Anissa Widjaja in the press release. She was testing a method to accurately detect IL-11. Several samples of an old rat’s proteins were in the mix, and she realized that IL-11 levels were far higher in the samples than in those from younger mice.

“From the readings, we could clearly see that the levels of IL-11 increased with age, and that’s when we got really excited,” she said.

Longevity Blocker

The results spurred the team to shift their research focus to longevity. A series of tests confirmed IL-11 levels consistently rose in a variety of tissues—muscle, fat, and liver—in both male and female mice as they aged.

To see how IL-11 influences the body, the team next deleted the gene coding for IL-11 and compared mice without the protein to their normal peers. At two years old, considered elderly for mice, tissues in normal individuals were littered with genetic signatures suggesting senescence—when cells lose their function but are still alive. Often called “zombie cells,” they spew out a toxic mix of inflammatory molecules and harm their neighbors. Elderly mice without IL-11, however, had senescence genetic profiles similar to those of much younger mice.

Deleting IL-11 had other perks. Weight gain is common with age, but without IL-11, the mice maintained their slim shape and had lower levels of fat, greater lean muscle mass, and shiny, full coats of fur. It’s not just about looks. Cholesterol levels and markers for liver damage were far lower than in normal peers. Aged mice without IL-11 were also spared shaking tremors—otherwise common in elderly mice—and could flexibly adjust their metabolism depending on the quantity of food they ate.

The benefits also showed up in their genetic material. DNA is protected by telomeres—a sort of end cap on chromosomes—that dwindle in length with age. Ridding cells of IL-11 prevented telomeres from eroding away in the livers and muscles of the elderly mice.

Genetically deleting IL-11 is a stretch for clinical use in humans. The team next turned to a more feasible alternative: An antibody shot. Antibodies can grab onto a target, in this case IL-11, and prevent it from functioning.

Beginning at 75 weeks, roughly the equivalent of 55 human years, the mice received an antibody shot every month for 25 weeks—over half a year. Similar antibodies are already being tested in clinical trials.

The health benefits in these mice matched those in mice without IL-11. Their weight and fat decreased, and they could better handle sugar. They also fought off signs of frailty as they aged, experiencing minimal tremors and problems with gait and maintaining higher metabolisms. Rather than wasting away, their muscles were even stronger than at the beginning of the study.

The treatment didn’t just increase healthspan. Monthly injections of the IL-11 antibody until natural death also increased lifespan in both male and female mice by up to 25 percent.

“These findings are very exciting. The treated mice had fewer cancers and were free from the usual signs of aging and frailty… In other words, the old mice receiving anti-IL-11 were healthier,” said Cook.

Although IL-11 antibody drugs are already in clinical trials, translating these results to humans could face hurdles. Mice have a relatively short lifespan. A longevity trial in humans would be long and very expensive. The treated mice were also contained in a lab setting, whereas in the real world we roam around and have differing lifestyles—diet, exercise, drinking, smoking—that could confound results. Even if it works in humans, a shot every month beginning in middle age would likely rack up a hefty bill, providing health and life extension only to those who could afford it.

To Cook, rather than focusing on extending longevity per se, tackling a specific age-related problem, such as tissue scarring or losing muscles is a better alternative for now.

“While these findings are only in mice, it raises the tantalizing possibility that the drugs could have a similar effect in elderly humans. Anti-IL-11 treatments are currently in human clinical trials for other conditions, potentially providing exciting opportunities to study its effects in aging humans in the future,” he said.

Image Credit: MRC LMS, Duke-NUS Medical School

ARTIFICIAL INTELLIGENCE

The Data That Powers AI Is Disappearing Fast
Kevin Roose | The New York Times
“Over the past year, many of the most important web sources used for training AI models have restricted the use of their data, according to a study published this week by the Data Provenance Initiative, an MIT-led research group. The study, which looked at 14,000 web domains that are included in three commonly used AI training data sets, discovered an ’emerging crisis in consent,’ as publishers and online platforms have taken steps to prevent their data from being harvested.”

COMPUTING

How One Bad CrowdStrike Update Crashed the World’s Computers
Lily Hay Newman, Matt Burgess, and Andy Greenberg | Wired
“Only a handful of times in history has a single piece of code managed to instantly wreck computer systems worldwide. The Slammer worm of 2003. Russia’s Ukraine-targeted NotPetya cyberattack. North Korea’s self-spreading ransomware WannaCry. But the ongoing digital catastrophe that rocked the internet and IT infrastructure around the globe over the past 12 hours appears to have been triggered not by malicious code released by hackers, but by the software designed to stop them.”

ROBOTICS

Tiny Solar-Powered Drones Could Stay in the Air Forever
Matthew Sparkes | New Scientist
“A drone weighing just 4 grams is the smallest solar-powered aerial vehicle to fly yet, thanks to its unusual electrostatic motor and tiny solar panels that produce extremely high voltages. Although the hummingbird-sized prototype only operated for an hour, its makers say their approach could result in insect-sized drones that can stay in the air indefinitely.”

TECH

How Microsoft’s Satya Nadella Became Tech’s Steely Eyed AI Gambler
Karen Weise and Cade Metz | The New York Times
“Though it could be years before he knows if any of this truly pays off, Mr. Nadella sees the AI boom as an all-in moment for his company and the rest of the tech industry. He aims to make sure that Microsoft, which was slow to the dot-com boom and whiffed on smartphones, dominates this new technology.”

ENERGY

Chinese Nuclear Reactor Is Completely Meltdown-Proof
Alex Wilkins | New Scientist
“A large-scale nuclear power station in China is the first in the world to be completely impervious to dangerous meltdowns, even during a full loss of external power. …To test this [capability in the power station], which became commercially operational in December 2023, [Zhe] Dong and his team switched off both modules of HTR-PM as they were operating at full power, then measured and tracked how the temperature of different parts of the plant went down afterwards. They found that HTR-PM naturally cooled and reached a stable temperature within 35 hours after the power was removed.”

AUTOMATION

The AI-Powered Future of Coding Is Near
Will Knight | Wired
“I am by no means a skilled coder, but thanks to a free program called SWE-agent, I was just able to debug and fix a gnarly problem involving a misnamed file within different code repositories on the software-hosting site GitHub. I pointed SWE-agent at an issue on GitHub and watched as it went through the code and reasoned about what might be wrong. It correctly determined that the root cause of the bug was a line that pointed to the wrong location for a file, then navigated through the project, located the file, and amended the code so that everything ran properly.”

ENVIRONMENT

Balloons Will Surf Wind Currents to Track Wildfires
Sarah Scoles | MIT Technology Review
“Urban Sky aims to combine the advantages of satellites and aircraft by using relatively inexpensive high-altitude balloons that can fly above the fray—out of the way of airspace restrictions, other aircraft, and the fire itself. The system doesn’t put a human pilot at risk and has an infrared sensor system called HotSpot that provides a sharp, real-time picture, with pixels 3.5 meters across.”

ARTIFICIAL INTELLIGENCE

Here’s the Real Reason AI Companies Are Slimming Down Their Models
Mark Sullivan | Fast Company
“OpenAI is one of a number of AI companies to develop a version of its best ‘foundation’ model that trades away some intelligence for some speed and affordability. Such a trade-off could let more developers power their apps with AI, and may open the door for more complex apps like autonomous agents in the future.”

SPACE

Will Space-Based Solar Power Ever Make Sense?
Kat Friedrich | Ars Technica
“Is space-based solar power a costly, risky pipe dream? Or is it a viable way to combat climate change? Although beaming solar power from space to Earth could ultimately involve transmitting gigawatts, the process could be made surprisingly safe and cost-effective, according to experts from Space Solar, the European Space Agency, and the University of Glasgow. But we’re going to need to move well beyond demonstration hardware and solve a number of engineering challenges if we want to develop that potential.”

Image Credit: Edward Chou / Unsplash

Despite their uncanny language skills, today’s leading AI chatbots still struggle with reasoning. A secretive new project from OpenAI could reportedly be on the verge of changing that.

While today’s large language models can already carry out a host of useful tasks, they’re still a long way from replicating the kind of problem-solving capabilities humans have. In particular, they’re not good at dealing with challenges that require them to take multiple steps to reach a solution.

Imbuing AI with those kinds of skills would greatly increase its utility and has been a major focus for many of the leading research labs. According to recent reports, OpenAI may be close to a breakthrough in this area.

An article in Reuters claimed its journalists had been shown an internal document from the company discussing a project code-named Strawberry that is building models capable of planning, navigating the internet autonomously, and carrying out what OpenAI refers to as “deep research.”

A separate story from Bloomberg said the company had demoed research at a recent all-hands meeting that gave its GPT-4 model skills described as similar to human reasoning abilities. It’s unclear whether the demo was part of project Strawberry.

According, to the Reuters report, project Strawberry is an extension of the Q* project that was revealed last year just before OpenAI CEO Sam Altman was ousted by the board. The model in question was supposedly capable of solving grade-school math problems.

That might sound innocuous, but some inside the company believed it signaled a breakthrough in problem-solving capabilities that could accelerate progress towards artificial general intelligence, or AGI. Math has long been an Achilles’ heel for large language models, and capabilities in this area are seen as a good proxy for reasoning skills.

A source told Reuters that OpenAI has tested a model internally that achieved a 90 percent score on a challenging test of AI math skills, though it again couldn’t confirm if this was related to project Strawberry. But another two sources reported seeing demos from the Q* project that involved models solving math and science questions that would be beyond today’s leading commercial AIs.

Exactly how OpenAI has achieved these enhanced capabilities is unclear at present. The Reuters report notes that Strawberry involves fine-tuning OpenAI’s existing large language models, which have already been trained on reams of data. The approach, according to the article, is similar to one detailed in a 2022 paper from Stanford researchers called Self-Taught Reasoner or STaR.

That method builds on a concept known as “chain-of-thought” prompting, in which a large language model is asked to explain the reasoning steps behind its answer to a query. In the STaR paper, the authors showed an AI model a handful of these “chain-of-thought” rationales as examples and then asked it to come up with answers and rationales for a large number of questions.

If it got the question wrong, the researchers would show the model the correct answer and then ask it to come up with a new rationale. The model was then fine-tuned on all of the rationales that led to a correct answer, and the process was repeated. This led to significantly improved performance on multiple datasets, and the researchers note that the approach effectively allowed the model to self-improve by training on reasoning data it had produced itself.

How closely Strawberry mimics this approach is unclear, but if it relies on self-generated data, that could be significant. The holy grail for many AI researchers is “recursive self-improvement,” in which weak AI can enhance its own capabilities to bootstrap itself to higher orders of intelligence.

However, it’s important to take vague leaks from commercial AI research labs with a pinch of salt. These companies are highly motivated to give the appearance of rapid progress behind the scenes.

The fact that project Strawberry seems to be little more than a rebranding of Q*, which was first reported over six months ago, should give pause. As far as concrete results go, publicly demonstrated progress has been fairly incremental, with the most recent AI releases from OpenAI, Google, and Anthropic providing modest improvements over previous versions.

At the same time, it would be unwise to discount the possibility of a significant breakthrough. Leading AI companies have been pouring billions of dollars into making the next great leap in performance, and reasoning has been an obvious bottleneck on which to focus resources. If OpenAI has genuinely made a significant advance, it probably won’t be long until we find out.

Image Credit: gemenu / Pixabay

Magic mushrooms have recently had a reputation revamp. Often considered a hippie drug, their main active component, psilocybin, is being tested in a variety of clinical trials as a therapy for the likes of depression and post-traumatic stress, bipolar, and eating disorders.

Psilocybin joins ketamine, LSD (commonly known as acid), and MDMA (often called ecstasy or molly) as part of the psychedelic therapy renaissance. But the field has had some ups and downs.

In 2019, the FDA approved a type of ketamine for severe depression that was resistant to other therapies. Then in early June, the agency rejected MDMA therapy for post-traumatic stress disorder, although it has been approved for limited use in Australia. Meanwhile, healthcare practitioners in Oregon are already using psilocybin, in combination with counseling, to treat depression, although the drug hasn’t yet been federally approved.

Despite its potential, no one knows how psilocybin works in the brain, especially over longer durations.

Now, a team from Washington University School of Medicine has comprehensively documented brain-wide changes before, during, and after a single dose of psilocybin over a period of weeks. As a control, the volunteers also took Ritalin, a stimulant, at a different time to mimic parts of the psilocybin high.

An fMRI scan shows the effect of psilocybin on the brain. Yellows, oranges, and reds indicate an increasingly large departure from normal activity. Image Credit: Sara Moser/Washington University

In the study, psilocybin dramatically reset brain networks that hum along during active rest—say, while daydreaming or spacing out. These networks control our sense of self, time, and space. Although most effects were temporary, one connection showed changes for weeks.

In some participants, the alterations were so drastic that their brain connections resembled those of completely different people.

Normally, the brain synchronizes activity across regions. Psilocybin disrupts these connections, in turn making the brain more malleable and ready to form new networks.

This could be how magic mushrooms “contribute to persistent changes…in brain regions that are responsible for controlling a person’s sense of self, emotion, and life-narrative,” wrote Petros Petridis at the NYU Langone Center for Psychedelic Medicine, who was not involved in the study.

Magical Mystery Tour

The brain’s 100 billion neurons and trillions of connections are highly organized into local and brain-wide networks.

Local networks tackle immediate tasks such as processing vision, sound, or motor functions. Brain-wide networks integrate information from local networks to coordinate more complex tasks, such as decision-making, reasoning, or self-reflection.

Previous psilocybin studies mainly focused on local networks. In rodents, for example, the drug regrew neural connections that often wither away in people with severe depression. Scientists have also pinpointed a receptor—which psilocybin grabs onto—that triggers this growth.

But psilocybin’s effects on the whole human brain remained a mystery.

Several years back, one team sought an answer by giving people with severe depression a dose of psilocybin. Using functional MRI (fMRI), a type of imaging that captures brain activity based on changes in blood flow, they found the chemical desynchronized neural networks across the entire brain, essentially “rebooting” them out of a depressive state.

Daydream Believer

The new study used fMRI to track brain activity in seven adults without mental health struggles before, during, and for three weeks after they took psilocybin. The researchers gave participants a single dose on par with that commonly used in clinical trials for depression.

During the scans, the participants had two tasks. One sounds easy: They kept still and focused their gaze on white crosshairs on a computer screen, but remained otherwise relatively relaxed. Even so, tripping on mushrooms inside a noisy, claustrophobic machine is hardly relaxing—heart rate skyrockets, nerves are on high alert, and anxiety rapidly builds. To control for these side effects, the participants also took Ritalin—a stimulant commonly used to manage attention deficit hyperactivity disorder—at another point in time during the study.

The other task required more brain power. Like an audio version of a CAPTCHA, the researchers asked volunteers to match an image and a word prompt—for example, they’d have to pick a photo of a beach after hearing the word “beach.”

Throughout the study, each person had their brains scanned roughly every other day, on average totaling 18 scans.

Mapping brain connections over time in the same person can “minimize the effects of individual differences in brain networks organization,” wrote Petridis.

The study found psilocybin immediately desynchronized a brain-wide network, generating a brain activation “fingerprint” of sorts that differentiates it from a sober brain.

Dubbed the default mode network, this neural system is active when the mind is alert but wanders, like when reliving previous memories or imagining future scenarios. The network is distributed across the brain and is often studied for its role in consciousness and a sense of self. The chemical also desynchronized local networks across the cortex, the outermost layer of the brain that supports perception, reasoning, and decision-making.

However, the chemical partially lost its magic when the volunteers were focused on the image-audio task, at which point the scans showed less disruption to the default mode network.

This has implications for psilocybin-assisted treatment. Clinical studies have shown that during psychedelic therapy, a challenging experience—a bad trip—can be overcome by a method called “grounding,” which reconnects the person to the outside world.

These results could explain why adding eye masks and ear plugs can enhance the therapeutic experience by blocking outside stimulation, while grounding pulls one out of a bad trip.

Psilocybin’s effects lingered for a few days, after which most brain networks returned to normal—with one exception. A link between the default mode network and a part of the brain involved in creating memories, emotions, and a sense of time, space, and self was disrupted for weeks.

In a way, psilocybin opens a window during which neural connections become more malleable and easier to rewire. People with depression or post-traumatic stress disorder often have a rigid and maladaptive thought pattern that’s hard to shake off. With therapy, psilocybin allows the brain to reorganize those networks, potentially helping people with depression to escape negative ruminations or for people suffering from addiction to consider a new perspective on their relationship to substances.

“In other words, psilocybin could open the door to change, allowing the therapist to lead the patient through,” wrote Petridis.

Although the study offered a higher resolution image of the brain on mushrooms over a longer timeframe than ever before, it only captured scans of seven people. As the participants did not have mental health issues, their responses to psilocybin may differ from those most likely to benefit therapeutically.

Ultimately, larger studies in diverse patient populations—as in several recent MDMA trials—could offer more insights into the efficacy of psilocybin therapy. For example, the one persistent brain network disruption could be an indicator of treatment efficacy. Investigating whether other psychedelics alter the same neural connection is a worthy next step, wrote Petridis.

With the field of psychedelic therapy projected to reach over $10 billion by 2027, understanding how the drug affects the brain could bring new medications with fewer side effects.

Image Credit: Sara Moser/Washington University

Is it possible that one day we could make Mars like Earth? –Tyla, age 16, Mississippi

When I was in middle school, my biology teacher showed our class the sci-fi movie Star Trek III: The Search for Spock.

The plot drew me in with its depiction of the “Genesis Project”—a new technology that transformed a dead alien world into one brimming with life.

After watching the movie, my teacher asked us to write an essay about such technology. Was it realistic? Was it ethical? And to channel our inner Spock: Was it logical? This assignment had a huge impact on me.

Fast-forward to today, and I’m an engineer and professor developing technologies to extend the human presence beyond Earth.

For example: I’m working on advanced propulsion systems to take spacecraft beyond Earth’s orbit. I’m helping to develop lunar construction technologies to support NASA’s goal of a long-term human presence on the moon. And I’ve been on a team that showed how to 3D print habitats on Mars.

To sustain people beyond Earth will take a lot of time, energy, and imagination. But engineers and scientists have started to chip away at the many challenges.

A photo taken of the bleak Martian surface by NASA’s Perseverance rover in June 2024. Image Credit: NASA/JPL-Caltech A Partial Checklist: Food, Water, Shelter, Air

After the moon, the next logical place for humans to live beyond Earth is Mars.

But is it possible to terraform Mars—that is, transform it to resemble the Earth and support life? Or are these just the musings of science fiction?

To live on Mars, humans will need liquid water, food, shelter, and an atmosphere with enough oxygen to breathe and that’s thick enough to retain heat and protect against radiation from the sun.

But the Martian atmosphere is almost all carbon dioxide, with virtually no oxygen. And it’s very thin—only about 1 percent as dense as the Earth’s.

The less dense an atmosphere, the less heat it can hold onto. Earth’s atmosphere is thick enough to retain the heat needed to sustain life by what’s known as the greenhouse effect.

But on Mars, the atmosphere is so slight that the nighttime temperature drops routinely to -150 degrees Fahrenheit (-101 degrees Celsius).

So what’s the best way to give Mars an atmosphere?

Although Mars has no active volcanoes now—at least as far as we know—scientists could trigger volcanic eruptions via nuclear explosions. The gases trapped deep in a volcano would be released and then drift into the atmosphere. But that scheme is a bit harebrained because the explosions would also introduce deadly radioactive material into the air.

A better idea: Redirecting water-rich comets and asteroids to crash into Mars. That too would release gases from below the planet’s surface into the atmosphere while also releasing the water found in the comets. NASA has already demonstrated that it is possible to redirect asteroids—but relatively large ones, and lots of them, are needed to make a difference.

Making Mars Cozy

There are numerous ways to heat up the planet. For instance, gigantic mirrors, built in space and placed in orbit around Mars, could reflect sunlight to the surface and warm it up.

One recent study proposed that Mars colonists could spread aerogel, an ultralight solid material, on the ground. The aerogel would act as insulation and trap heat. This could be done all over Mars, including the polar ice caps, where the aerogel could melt the existing ice to make liquid water.

To grow food, you need soil. On Earth, soil is composed of five ingredients: minerals, organic matter, living organisms, gases, and water.

But Mars is covered in a blanket of loose, dust-like material called regolith. Think of it as Martian sand. The regolith contains few nutrients, not enough for healthy plant growth, and it hosts some nasty chemicals called perchlorates, used on Earth in fireworks and explosives.

Cleaning up the regolith and turning it into something viable wouldn’t be easy. What the alien soil needs is some Martian fertilizer, maybe made by adding extremophiles to it—hardy microbes imported from Earth that can survive even the harshest conditions. Genetically engineered organisms are also a possibility.

Through photosynthesis, these organisms would begin converting carbon dioxide to oxygen. Eventually, as Mars became more friendly to Earth-like organisms, colonists could introduce more complex plants and even animals.

Providing oxygen, water, and food in the right proportions is extraordinarily complex. On Earth, scientists have tried to simulate this in Biosphere 2, a closed-off ecosystem featuring ocean, tropical, and desert habitats. Although all of Biosphere 2’s environments are controlled, even there scientists struggle to get the balance right. Mother Nature really knows what she’s doing.

A House on Mars

Buildings could be 3D printed; initially, they would need to be pressurized and protected until Mars acquired Earth-like temperatures and air. NASA’s Moon-to-Mars Planetary Autonomous Construction Technologies program is researching how to do exactly this.

There are many more challenges. For example, unlike Earth, Mars has no magnetosphere, which protects a planet from solar wind and cosmic radiation. Without a magnetic field, too much radiation gets through for living things to stay healthy. There are ways to create a magnetic field, but so far the science is highly speculative.

In fact, all the technologies I’ve described are far beyond current capabilities at the scale needed to terraform Mars. Developing them would take enormous amounts of research and money, probably much more than possible in the near term. Although the Genesis device from Star Trek III could terraform a planet in a matter of minutes, terraforming Mars would take centuries or even millennia.

And there are a lot of ethical questions to resolve before people get started on turning Mars into another Earth. Is it right to make such drastic permanent changes to another planet?

If this all leaves you disappointed, don’t be. As scientists create innovations to terraform Mars, we’ll also use them to make life better on Earth. Remember the technology we’re developing to 3D print habitats on Mars? Right now, I’m part of a group of scientists and engineers employing that very same technology to print homes here on Earth—which will help address the world’s housing shortage.

Curious Kids is a series for children of all ages. If you have a question you’d like an expert to answer, send it to [email protected].

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Image Credit: Daein Ballard / Wikimedia Commons

With half their skulls replaced by translucent 3D-printed implants, the mice looked straight out of a science fiction movie. Yet they nosed around, ferociously ate their chow, and groomed as usual. Meanwhile, sensors spread across half their brains recorded electrical chatter.

Brain implants have revolutionized neuroscience. Our perception, thoughts, emotions, and memories all rely on electrical signals spreading across networks of neurons. Implants tap into these signals and—often with the help of AI—can rapidly decipher seemingly random electrical activity into intent or movement.

So-called “mind reading” devices translate brain signals related to speech into text, allowing people who’ve lost their ability to speak to communicate directly with loved ones with their minds. Others tap into motor regions of the brain or nerves in the spinal cord and help people with severe paralysis to walk again. Neural signals, alone or combined with eye movements, can even control cursors on a computer screen, re-opening the digital world to paralyzed people for texting, Googling, and scrolling through social media.

These devices are beginning to transform lives for the better. But all rely on the answer to one critical question: How does the brain support those functions? So far, each implant has focused on a small brain region underlying a given capability—controlling vision or movement.

But many of the brain’s functions rely on signals, or brain waves, that spread across multiple regions and synchronize electrical activity. The height and frequency of the waves—some come fast and low, others slow and high—change the brain’s overall function.

Scientists can measure these waves, but like a camera with low resolution, they can’t explain how the waves are generated, propagate, and eventually die down.

In a new study, the custom-fitted transparent implant described above replaces the skull in mice, offering a way to, literally, peek into the brain in search of answers.

The Evolution of Brain Probes

Neural implants have been around since the 1980s. The idea is simple. The brain uses electrical and chemical signals to process information. Electrodes can tap into the electrical communications. Sophisticated software then deciphers the neural code, potentially allowing us to reprogram it and tackle neurological symptoms when the code breaks down.

There are a few ways to make it work. One is to directly record individual neurons—often in rodents—to see which activate when challenging a mouse to a task. Another technology records large-scale brain activity from beneath the skull. This approach sacrifices resolution—we no longer know how each individual neuron behaves—but paints a broader picture.

The challenge is how to combine resolution and scale. A previous attempt relied on multiple high-density electrodes inserted into the brain. Called Neuropixels, each implant is a powerhouse with over 5,000 recording sites packed in a tiny, durable package. “Extremely large numbers of individual neurons could…be followed and tracked with the same probe for weeks and occasionally months,” the authors of a paper about the implant wrote at the time.

But to measure brain-wide activity, scientists have to place multiple Neuropixel devices across the brain. Each requires drilling through the skull and could harm the plastic-wrap-like structure, known as the blood-brain barrier, that protects the brain. Damage from these surgeries often compounds, triggering inflammation that could change how the brain works for weeks with an increasing risk of infection.

So far, scientists have inserted up to eight implants to record activity in mice as they went about their lives or participated in experiments. While they gained news insights, scientists struggled to keep the mice healthy after multiple surgeries. In other words, it wasn’t the Neuropixel implants causing problems—it was all the brain surgeries.

Might there be an alternative?

A 3D Replacement

To avoid multiple surgeries, the Allen Institute team behind the new study developed an implant that covers nearly half a mouse’s brain.

Called SHIELD, the implant looks a bit like molded Swiss cheese. The scaffold is carefully contoured into a shape that perfectly mimics the skulls of young mice.

Then comes the customization. The SHIELD scaffold can accommodate up to 21 small insertion holes for Neuropixels. Scientists can strategically choose where to put the holes to record from multiple brain regions of interest. The implant is then printed with resin, a viscous liquid often used in everyday 3D printing.

“The SHIELD implant is straightforward to fabricate in-house, using a commercially available and relatively low-cost 3D printer,” wrote the team.

Once printed, each hole is temporarily filled with transparent silicon rubber. Like a flexible windshield, the rubber protects delicate brain tissue during implantation. The SHIELD then replaces half of the skull in a single surgery.

It sounds traumatic, but the team made sure the procedure didn’t harm the mice’s health or brains. Images of their brains at multiple time points over two months after surgery showed little damage in most mice, who went about their merry business after a short recovery period. Brain inflammation levels also stayed low during the study.

Here’s how it went. The mice watched one of eight photos flash before them continuously and then learned to lick a treat when the photo switched. During the test, six Neuropixel implants recorded brain activity related to the task. The position of the implants changed every day for four days, altogether collecting neural signals from roughly 25 different brain areas.

Another test dug into the potential underpinnings of brain waves first discovered in the 1920s. These neural oscillations, called alpha waves, are associated with restful and meditative states. In humans, brain waves are usually monitored using a beanie-like cap covered in electrodes that can record most brain regions. With the help of AI, Neuropixel recordings from across the brain homed in on signals resembling alpha waves.

Overall, the team made stable, high-quality recordings from 25 mice, with 467 probe insertions across nearly 90 different experiments.

“Thus, this work goes beyond mere proof of concept,” instead providing a solid recipe for recording across dozens of brain regions and thousands of neurons over multiple days with a single initial surgery, wrote the team.

And there’s a final perk. Because SHIELD is translucent, scientists can tweak brain activity using light. This approach, called optogenetics, alters brain activity with flashes of light—either amping it up or turning it down—giving scientists insight into the neural underpinnings of thoughts, emotions, and memories.

The authors shared 3D printing files of the scaffold for other scientists to design their own custom implants.

Image Credit: Ralph / Pixabay

A new generation of “flying cars” promises to revolutionize urban mobility, but limited battery power holds them back from plying longer routes. A new hydrogen-powered variant from Joby Aviation could soon change that.

Rapid advances in battery technology and electric motors have opened the door to a new class of aircraft known as eVTOLs, which stands for electric vertical takeoff and landing. The companies making the aircraft tout them as a quieter, greener alternative to helicopters.

However, current battery technology means they’re limited to ranges of approximately 150 miles. That’s why they have primarily been envisaged as a new form of urban mobility, allowing quick hops across cities congested with traffic.

Joby is already developing a battery-powered eVTOL that it expects to start commercial operations next year. But this week, the company announced it has created a hydrogen-powered version of the aircraft, which recently completed a 523-mile test flight. The company says this could allow eVTOLs to break into regional travel as well.

“With our battery-electric air taxi set to fundamentally change the way we move around cities, we’re excited to now be building a technology stack that could redefine regional travel using hydrogen-electric aircraft,” JoeBen Bevirt, founder and CEO of Joby, said in a press release.

“Imagine being able to fly from San Francisco to San Diego, Boston to Baltimore, or Nashville to New Orleans without the need to go to an airport and with no emissions except water.”

Joby’s demonstrator is a converted battery-electric aircraft that had already completed 25,000 miles of test flights. It features the same airframe with six electric-motor-powered tilting propellers that allow it to take off vertically like a helicopter but cruise like a light aircraft. Joby says this should significantly speed up the certification process if the company decides to commercialize the technology.

What’s new is the addition of a hydrogen fuel cell system designed by H2FLY, a German startup Joby acquired in 2021, and a liquid hydrogen fuel tank that can store about 40 kilograms of fuel. The fuel cell combines the liquid hydrogen with oxygen from the air to generate the electricity that powers the aircraft’s motors. The H2FLY team used the same underlying technology in a series of demonstration flights with a more conventional aircraft design last year.

The new Joby aircraft will still carry some batteries to provide additional power during takeoff and landing. But hydrogen has a much higher energy density—or specific energy—than batteries, which makes it possible to power the aircraft for significantly longer.

“Hydrogen has one hundred times the specific energy of today’s batteries and three times that of jet fuel,” Bevirt wrote in a blog post. “The result is an electric aircraft that can travel much farther—and carry a greater payload—than is possible not only with any battery cells currently under development, but even with the same mass of jet fuel.”

However, switching to hydrogen fuel poses some challenges. For a start, hydrogen requires complicated cooling equipment, which means airports or other landing facilities would need to invest significant amounts in new fueling infrastructure.

“The industry is already scratching its head figuring out how to support battery electric aircraft with charging infrastructure at airports,” Cyrus Sigari, co-founder and managing partner of VC Up.Partners, told TechCrunch. “Adding hydrogen filling stations into that equation will present even more challenges.”

Hydrogen’s green credentials are also somewhat weaker than those of batteries. While it’s possible to generate hydrogen from water using only renewable electricity, at present the vast majority is produced from fossil fuels.

However, efforts are underway to increase the supply of green hydrogen, and the Bipartisan Infrastructure Law passed in 2021 set aside $9.5 billion to help boost these efforts. And if hydrogen-powered flight can piggyback on innovations in eVTOL technology, it could prove a powerful way to curb emissions in one of the world’s most polluting sectors.

Image Credit: Joby

ARTIFICIAL INTELLIGENCE

OpenAI Reportedly Nears Breakthrough With ‘Reasoning’ AI, Reveals Progress Framework
Benj Edwards | Ars Technica
“[According to OpenAI’s new AGI framework] a Level 2 AI system would reportedly be capable of basic problem-solving on par with a human who holds a doctorate degree but lacks access to external tools. During the all-hands meeting, OpenAI leadership reportedly demonstrated a research project using their GPT-4 model that the researchers believe shows signs of approaching this human-like reasoning ability, according to someone familiar with the discussion who spoke with Bloomberg.”

BIOTECH

How AI Revolutionized Protein Science, but Didn’t End It
Yasemin Saplakoglu | Quanta
“Three years ago, Google’s AlphaFold pulled off the biggest artificial intelligence breakthrough in science to date, accelerating molecular research, and kindling deep questions about why we do science. …’The field of protein biology is ‘more exciting right now than it was before AlphaFold,’ Perrakis said. The excitement comes from the promise of reviving structure-based drug discovery, the acceleration in creating hypotheses, and the hope of understanding complex interactions happening within cells.”

TECH

New Fiber Optics Tech Smashes Data Rate Record 
Margo Anderson | IEEE Spectrum
“An international team of researchers have smashed the world record for fiber optic communications through commercial-grade fiber. By broadening fiber’s communication bandwidth, the team has produced data rates four times as fast as existing commercial systems—and 33 percent better than the previous world record.”

ARTIFICIAL INTELLIGENCE

‘Superhuman’ Go AIs Still Have Trouble Defending Against These Simple Exploits
Kyle Orland | Ars Technica
“In the ancient Chinese game of Go, state-of-the-art artificial intelligence has generally been able to defeat the best human players since at least 2016. But in the last few years, researchers have discovered flaws in these top-level AI Go algorithms that give humans a fighting chance. By using unorthodox ‘cyclic’ strategies—ones that even a beginning human player could detect and defeat—a crafty human can often exploit gaps in a top-level AI’s strategy and fool the algorithm into a loss.”

COMPUTING

Google Creates Self-Replicating Life From Digital ‘Primordial Soup’
Matthew Sparkes | New Scientist
“A self-replicating form of artificial life has arisen from a digital ‘primordial soup’ of random data, despite a lack of explicit rules or goals to encourage such behavior. Researchers believe it is possible that more sophisticated versions of the experiment could yield more advanced digital organisms, and if they did, the findings could shed light on the mechanisms behind the emergence of biological life on Earth.”

AUTOMATION

How Good Is ChatGPT at Coding, Really?
Michelle Hampson | IEEE Spectrum
“Programmers have spent decades writing code for AI models, and now, in a full circle moment, AI is being used to write code. But how does an AI code generator compare to a human programmer? [A new study shows] that ChatGPT has an extremely broad range of success when it comes to producing functional code—with a success rate ranging from anywhere as poor as 0.66 percent and as good as 89 percent—depending on the difficulty of the task, the programming language, and a number of other factors.”

TECH

OpenAI Anticipates Decrease in AI Model Costs Amid Adoption Surge
Shubham Sharma | VentureBeat
“‘We introduced GPT-4, the first version, some 15 months ago. Since then, the cost of a token/word on the model has been reduced by 85-90%. There’s no reason why that trend will not continue,’ Olivier Godement, [OpenAI’s] head of API Product said. …He expects the company’s work on affordability, spanning efforts to optimize costs at both hardware and inference levels, will continue, leading to a further decline in the cost of running frontier AI models—much like what has been the case with smartphones and televisions.”

SPACE

Watch These Supernovas in (Time-Lapse) Motion
Dennis Overbye | The New York Times
“This spring, the astronomers who operate Chandra combined its X-ray images into videos that document the evolution of two astrophysical landmarks: the Crab nebula, in the constellation Taurus, and Cassiopeia A, a gas bubble and hub of radio noise in the constellation Cassiopeia. The videos show twisting, drifting ribbons of the remains of the star being churned by shock waves and illuminated by radiation from the dense, spinning cores left behind.”

Image Credit: BoliviaInteligente / Unsplash

CRISPR was one of the most influential breakthroughs of the last decade, but it’s still imperfect. While the gene editing tool is already helping people with genetic ailments, scientists are also looking to improve on it.

Efforts have extended the CRISPR family to include less damaging, more accurate, and smaller versions of the gene editor. But in the bacterial world, where CRISPR was originally discovered, we’re only scratching the surface. Two new papers suggest an even more powerful gene editor may be around the corner—if it’s proven to work in cells like our own.

In one of the papers, scientists at the Arc Institute say they discovered a new CRISPR-like gene editing tool in bacterial “jumping genes.” Another paper, written independently, covers the same tool and extends the work to a similar one in a different family.

Jumping genes move around within genomes and even between individuals. It’s long been known they do this by cutting and pasting their own DNA, but none of the machinery has been shown to be programmable like CRISPR. In the recent studies, scientists describe jumping gene systems that, in a process the teams are alternatively calling bridge editing and seekRNA, can be modified to cut, paste, and flip any DNA sequence.

Crucially, unlike CRISPR, the system does all this without breaking strands of DNA or relying on the cell to repair them, a process that can be damaging and unpredictable. The various molecules involved are also fewer and smaller than those in CRISPR, potentially making the tool safer and easier to deliver into cells, and can deal with much longer sequences.

“Bridge recombination can universally modify genetic material through sequence-specific insertion, excision, inversion, and more, enabling a word processor for the living genome beyond CRISPR,” said Berkeley’s Patrick Hsu, a senior author of one of the studies and Arc Institute core investigator, in a press release.

CRISPR Coup

Scientists first discovered CRISPR in bacteria defending themselves against viruses. In nature, a Cas9 protein pairs with an RNA guide molecule to seek out viral DNA and, when located, chop it up. Researchers learned to reengineer this system to seek out any DNA sequence, including sequences found in human genomes, and break the DNA strands at those locations. The natural machinery of the cell then repairs these breaks, sometimes using a provided strand of DNA.

CRISPR gene editing is powerful. It’s being investigated in clinical trials as a treatment for a variety of genetic diseases and, late last year, received its first clinical approval as a therapy for sickle cell disease and beta thalassemia. But it’s not perfect.

Because the system breaks DNA and relies on the cell to repair these breaks, it can be imprecise and unpredictable. The tool also works primarily on short sections of DNA. While many genetic illnesses are due to point mutations, where a single DNA “letter” has been changed, the ability to work with longer sequences would broaden the technology’s potential uses in both synthetic biology and gene therapy.

Scientists have developed new CRISPR-based systems over the years to address these shortcomings. Some systems only break a single DNA strand or swap out single genetic “letters” to increase precision. Studies are also looking for more CRISPR-like systems by screening the whole bacterial universe; others have found naturally occurring systems in eukaryotic cells like our own.

The new work extends the quest by adding jumping genes into the mix.

An RNA Bridge

Jumping genes are a fascinating feat of genetic magic. These sequences of DNA can move between locations in the genome using machinery to cut and paste themselves. In bacteria, they even move between individuals. This sharing of genes could be one way bacteria acquire antibiotic resistance—one cell that’s evolved to evade a drug can share its genetic defenses with a whole population.

In the Arc Institute study, researchers looked into a specific jumping gene in bacteria called IS110. They found that when the gene is on the move, it calls a sequence of RNA—like the RNA guide in CRISPR—to facilitate the process. The RNA includes two loops: One binds the gene itself and the other seeks out and binds to the gene’s destination in the genome. It acts like a bridge between the DNA sequence and the specific location where it’s to be inserted. In contrast to CRISPR, once found, the sequence can be added without breaking DNA.

“Bridge editing [cuts and pastes DNA] in a single-step mechanism that recombines and re-ligates the DNA, leaving it fully intact,” Hsu told Fierce Biotech in an email. “This is very distinct from CRISPR editing, which creates exposed DNA breaks that require DNA repair and have been shown to create undesired DNA damage responses.”

Crucially, the researchers discovered both loops of RNA can be reprogrammed. That means scientists can specify a genomic location as well as what sequence should go there. In theory, the system could be used to swap in long genes or even multiple genes. As a proof of concept in E. coli bacteria, the team programmed IS110 to insert a DNA sequence almost 5,000 bases long. They also cut and inverted another sequence of DNA.

The study was joined by a different paper written independently by another team of scientists at the University of Sydney detailing both IS110 and a related enzyme in a different family, IS111, that they say is similarly programmable. In their paper, they called these systems “seekRNA.”

The tools rely on a single protein half the size of those in CRISPR. That means it may be easier to package them in harmless viruses or lipid nanoparticles—these are also used in Covid vaccines—and ferry them into cells where they can get to work.

The Next Jump

The approach has big potential, but there’s also a big caveat. So far, the researchers have  only shown it works in bacteria. CRISPR, on the other hand, is incredibly versatile, having proved itself in myriad cell types. Next, they hope to hone the approach further and adapt it to mammalian cells like ours. That may not be easy. The University of Tokyo’s Hiroshi Nishimasu says the IS110 family hasn’t yet shown itself amenable to such a task.

All this is to say it’s still early in the technology’s arc. Scientists knew about CRISPR years before they showed it was programmable, and it wasn’t put to work in human cells until 2013. Although it’s moved relatively quickly from lab to clinic since then, the first CRISPR-based treatments took years more to materialize.

At the least, the new work shows we haven’t exhausted all nature has to offer gene editing. The tech could also be useful in the realm of synthetic biology, where single cells are being engineered on grand scales to learn how life works at its most basic and how we might reengineer it. And if the new system can be adapted for human cells, it would be a useful new option in the development of safer, more powerful gene therapies.

“If this works in other cells, it will be game-changing,” Sandro Fernandes Ataide, a structural biologist at the University of Sydney and author on the paper detailing IS111 told Nature. “It’s opening a new field in gene editing.”

Image Credit: The Arc Institute

Roughly 52,000 years ago, a woolly mammoth died in the Siberian tundra. As her body flash froze in the biting cold, something remarkable happened: Her DNA turned into a fossil. It wasn’t only genetic letters that were memorialized—the cold preserved their intricate structure too.

Fast forward to 2018, when an international expedition to the area found her preserved body. The team took little bits of skin from her head and ear, hairs still intact.

From these samples, scientists built a three-dimensional reconstruction of a woolly mammoth’s genome down to the nanometer. The results were published in Cell today.

Like humans, the mammoth’s DNA strands are tightly packed into chromosomes inside cells. These sophisticated structures are hard to analyze in detail, even for humans, but they contain insights into which genes are turned on or off and how they’re organized in different cell types.

Previous attempts to reconstruct ancient DNA only had tiny snippets of genetic sequences. Like trying to put together a puzzle with missing pieces, the resulting DNA maps were incomplete.

Thanks to the newly discovered flash-frozen DNA, this mammoth project—pun intended—is the first to assemble an enormous ancient genome in 3D.

“This is a new type of fossil, and its scale dwarfs that of individual ancient DNA fragments—a million times more sequence,” said study author Erez Lieberman Aiden at Baylor College of Medicine in a statement.

Aiden’s team heavily collaborated with Love Dalén at the Center of Palaeogenetics in Sweden. In a separate study, Dalén’s team analyzed 21 Siberian woolly mammoth genomes and charted how the species survived for six millennia after a potentially catastrophic genetic “bottleneck.”

The mammoth genomes weren’t that different than those of today’s Asian and African elephants. All have 28 pairs of chromosomes, and their X chromosomes twist into unique structures unlike most mammals. Digging deeper, the team found genes that were turned on or off in the mammoth compared to its elephant cousins.

“Our analyses uncover new biology,” wrote Aiden’s team in their paper.

DNA Serendipity

Ancient DNA is hard to come by, but it offers invaluable clues about the evolutionary past. In the 1980s, scientists eager to probe genetic history showed ancient DNA, however fragmented, could be extracted and sequenced in samples from an extinct member of the horse family and Egyptian mummies.

Thanks to modern DNA sequencing, the study of ancient DNA “has subsequently undergone a remarkable expansion,” wrote Aiden’s team. It’s now possible to sequence whole genomes from extinct humans, animals, plants, and even pathogens spanning a million years.

Making sense of the fragments is another matter. One way to decipher ancient genetic codes is to compare them to the genomes of their closest living cousins, such as woolly mammoths and elephants. This way, scientists can figure out which parts of the DNA sequence remained unchanged and where evolution swapped letters or small fragments.

These analyses can link genetic changes to function, such as identifying which genes made mammoths woolly. But they can’t capture large-scale differences at the chromosomal level. Because DNA relies on the chromosome’s 3D structure to function, sequencing its letters alone misses valuable information, such as when and where genes are turned on or off.

Chromosome Puzzle Master

Enter Hi-C. Developed in 2009 to reconstruct human genomes, the technique detects interactions between different genetic sites inside the cell’s nucleus.

Here’s roughly how it works. DNA strands are like ribbons that twirl around proteins in a structure resembling beads on a string. Because of this arrangement, different parts of the DNA strand are closer to each other in physical space. Hi-C “glues” together sections that are near one another and tags the pairs. Alongside modern DNA sequencing, the technique produces a catalog of DNA fragments that interact in physical space. Like a 3D puzzle, scientists can then put the pieces back together.

“Imagine you have a puzzle that has three billion pieces, but you don’t have the picture of the final puzzle to work from,” study author Marc A. Marti-Renom said in the press release. “Hi-C allows you to have an approximation of that picture before you start putting the puzzle pieces together.”

But Hi-C can be impossible to use in ancient samples because the surviving fragments are so short they’ve erased any chromosome shapes. They’ve literally withered away over time.

In the new study, the team developed a new technique, called PaleoHi-C, to analyze ancient DNA specifically.

Scientists immediately treated samples in the field to reduce contamination. They generated roughly 4.4 billion “pairs” of physically aligned DNA sequences—some interacting within a single chromosome, others between two. Overall, they painted a 3D snapshot of the woolly mammoth’s genetic material and how it looked inside cells with nanoscale detail.

In the new reconstructions, the team identified chromosome territories—certain chromosomes are located in different regions of the nucleus—alongside other quirks, such as loops that bring pairs of distant genomic sites into close physical proximity to alter gene expression. These patterns differed between cell types, suggesting it’s possible to learn which genes are active, not just for the mammoth but also compared to its closest living relative, the Asian elephant.

Roughly 820 genes differed between the two, with 425 active in the mammoth but not in elephants, and a similar number inactivated in one but not the other. One inactive mammoth gene that’s active in elephants has a human variant that is also shut down in the Nunavik Inuit, an indigenous people who thrive in the arctic. The gene “may be relevant for adaptation to a cold environment,” wrote the team.

Another inactive gene may explain how the woolly mammoth got its name. In humans and sheep, shutting down the same gene can result in excessive hair or wool growth.

“For the first time, we have a woolly mammoth tissue for which we know roughly which genes were switched on and which genes were off,” said Marti-Renom in the release. “This is an extraordinary new type of data, and it’s the first measure of cell-specific gene activity of the genes in any ancient DNA sample.”

Crystalized DNA

How did the mammoth’s genome architecture remain so well preserved for over 50,000 years?

Dehydration, often used to preserve food, may have been key. Using Hi-C on fresh beef, beef after 96 hours sitting on a desk, or jerky after a year at room temperature, the jerky took the win for resiliency. Even after getting run over by a car, immersed in acid, and pulverized by a shotgun (no joke), the dehydrated beef’s genomic architecture remained intact.

Dehydration could also partly be why the mammoth sample lasted so long. A chemical process called “glass transition” is widely used to produce shelf-stable food such as tortilla chips and instant coffee. It prevents pathogens from taking over or breaking down food. The mammoth’s DNA may also have been preserved in a glassy state called “chromoglass.” In other words, the sample was preserved across millennia by being freeze-dried.

It’s hard to say how long DNA architecture can survive as chromoglass, but the authors estimate it’s likely over two million years. Whether PaleoHi-C can work on hot-air-dried specimens, such as ancient Egyptian samples, remains to be seen.

As for mammoths, the next step is to examine gene expression patterns in other tissues and compare them to Asian elephants. Besides building an evolutionary throughline, the efforts could also guide ongoing studies looking to revive some version of the majestic animals.

“These results have obvious consequences for contemporary efforts aimed at woolly mammoth de-extinction,” said study author Thomas Gilbert at the University of Copenhagen in the release.

Image Credit: Beth Zaiken

Computer chips are a hot commodity. Nvidia is now one of the most valuable companies in the world, and the Taiwanese manufacturer of Nvidia’s chips, TSMC, has been called a geopolitical force. It should come as no surprise, then, that a growing number of hardware startups and established companies are looking to take a jewel or two from the crown.

Of these, Cerebras is one of the weirdest. The company makes computer chips the size of tortillas bristling with just under a million processors, each linked to its own local memory. The processors are small but lightning quick as they don’t shuttle information to and from shared memory located far away. And the connections between processors—which in most supercomputers require linking separate chips across room-sized machines—are quick too.

This means the chips are stellar for specific tasks. Recent preprint studies in two of these—one simulating molecules and the other training and running large language models—show the wafer-scale advantage can be formidable. The chips outperformed Frontier, the world’s top supercomputer, in the former. They also showed a stripped down AI model could use a third of the usual energy without sacrificing performance.

Molecular Matrix

The materials we make things with are crucial drivers of technology. They usher in new possibilities by breaking old limits in strength or heat resistance. Take fusion power. If researchers can make it work, the technology promises to be a new, clean source of energy. But liberating that energy requires materials to withstand extreme conditions.

Scientists use supercomputers to model how the metals lining fusion reactors might deal with the heat. These simulations zoom in on individual atoms and use the laws of physics to guide their motions and interactions at grand scales. Today’s supercomputers can model materials containing billions or even trillions of atoms with high precision.

But while the scale and quality of these simulations has progressed a lot over the years, their speed has stalled. Due to the way supercomputers are designed, they can only model so many interactions per second, and making the machines bigger only compounds the problem. This means the total length of molecular simulations has a hard practical limit.

Cerebras partnered with Sandia, Lawrence Livermore, and Los Alamos National Laboratories to see if a wafer-scale chip could speed things up.

The team assigned a single simulated atom to each processor. So they could quickly exchange information about their position, motion, and energy, the processors modeling atoms that would be physically close in the real world were neighbors on the chip too. Depending on their properties at any given time, atoms could hop between processors as they moved about.

The team modeled 800,000 atoms in three materials—copper, tungsten, and tantalum—that might be useful in fusion reactors. The results were pretty stunning, with simulations of tantalum yielding a 179-fold speedup over the Frontier supercomputer. That means the chip could crunch a year’s worth of work on a supercomputer into a few days and significantly extend the length of simulation from microseconds to milliseconds. It was also vastly more efficient at the task.

“I have been working in atomistic simulation of materials for more than 20 years. During that time, I have participated in massive improvements in both the size and accuracy of the simulations. However, despite all this, we have been unable to increase the actual simulation rate. The wall-clock time required to run simulations has barely budged in the last 15 years,” Aidan Thompson of Sandia National Laboratories said in a statement. “With the Cerebras Wafer-Scale Engine, we can all of a sudden drive at hypersonic speeds.”

Although the chip increases modeling speed, it can’t compete on scale. The number of simulated atoms is limited to the number of processors on the chip. Next steps include assigning multiple atoms to each processor and using new wafer-scale supercomputers that link 64 Cerebras systems together. The team estimates these machines could model as many as 40 million tantalum atoms at speeds similar to those in the study.

AI Light

While simulating the physical world could be a core competency for wafer-scale chips, they’ve always been focused on artificial intelligence. The latest AI models have grown exponentially, meaning the energy and cost of training and running them has exploded. Wafer-scale chips may be able to make AI more efficient.

In a separate study, researchers from Neural Magic and Cerebras worked to shrink the size of Meta’s 7-billion-parameter Llama language model. To do this, they made what’s called a “sparse” AI model where many of the algorithm’s parameters are set to zero. In theory, this means they can be skipped, making the algorithm smaller, faster, and more efficient. But today’s leading AI chips—called graphics processing units (or GPUs)—read algorithms in chunks, meaning they can’t skip every zeroed out parameter.

Because memory is distributed across a wafer-scale chip, it can read every parameter and skip zeroes wherever they occur. Even so, extremely sparse models don’t usually perform as well as dense models. But here, the team found a way to recover lost performance with a little extra training. Their model maintained performance—even with 70 percent of the parameters zeroed out. Running on a Cerebras chip, it sipped a meager 30 percent of the energy and ran in a third of the time of the full-sized model.

Wafer-Scale Wins?

While all this is impressive, Cerebras is still niche. Nvidia’s more conventional chips remain firmly in control of the market. At least for now, that appears unlikely to change. Companies have invested heavily in expertise and infrastructure built around Nvidia.

But wafer-scale may continue to prove itself in niche, but still crucial, applications in research. And it may be the approach becomes more common overall. The ability to make wafer-scale chips is only now being perfected. In a hint at what’s to come for the field as a whole, the biggest chipmaker in the world, TSMC, recently said it’s building out its wafer-scale capabilities. This could make the chips more common and capable.

For their part, the team behind the molecular modeling work say wafer-scale’s influence could be more dramatic. Like GPUs before them, adding wafer-scale chips to the supercomputing mix could yield some formidable machines in the future.

“Future work will focus on extending the strong-scaling efficiency demonstrated here to facility-level deployments, potentially leading to an even greater paradigm shift in the Top500 supercomputer list than that introduced by the GPU revolution,” the team wrote in their paper.

Image Credit: Cerebras

Henry Grabar has had enough battling knotweed. All he wanted was to build a small garden in Brooklyn—a bit of peace amid the cacophony of city life. But a plant with beet-red leaves soon took over his nascent garden. The fastest growing plant he’d ever seen, it could sprout up to 10 feet high and grow thick as a cornfield. Even with herbicide, it was nearly impossible to kill.

Invasive plant species and weeds don’t just ruin backyard gardens. Weeds decrease crop yields at an average annual cost of $33 billion, and control measures can rack up $6 billion more. Herbicides are a defense, but they have their own baggage. Weeds rapidly build resistance against the chemicals, and the resulting produce can be a hard sell for many consumers.

Weeds often seem to have the upper hand. Can we take it away?

Two recent studies say yes. Using a technology called a synthetic gene drive, the teams spliced genetic snippets into a mustard plant popular in lab studies. Previously validated in fruit flies, mosquitoes, and mice, gene drives break the rules of inheritance, allowing “selfish” genes to rapidly spread across entire species.

But making gene drives work in plants has been a headache, in part due to the way they repair their DNA. The new studies found a clever workaround, leading to roughly 99 percent propagation of a synthetic genetic payload to subsequent generations, in contrast to nature’s 50 percent. Computer models suggest the gene drives could spread throughout an entire population of the plant in roughly 10 to 30 generations.

Overriding natural evolution, gene drives could add genes that make weeds more vulnerable to herbicides or reduce their pollination and numbers. Beneficial genes can also spread across crops—essentially fast-tracking the practice of cross-breeding for desirable traits.

“Imagine a future where yield-robbing agricultural weeds or biodiversity threatening invasive plants could be kept on a genetic leash,” wrote Paul Neve at the University of Copenhagen and Luke Barrett at CSIRO Agriculture and Food in Australia, who were not involved in the study.

50/50

Inheritance is a coin toss for most species. Half of an offspring’s genetic material comes from each parent.

Gene drives torpedo this inheritance rule. Developed roughly a decade ago, the technology relies on CRISPR—the gene editing tool—to spread a new gene throughout a population, beating the 50/50 odds. In insects and mammals, a gene can propagate at roughly 80 percent, shuttling an inherited trait down generations and irreversibly changing an entire species.

While this may seem somewhat nefarious, gene drives are designed for good. A main use under investigation is to control disease-carrying mosquitoes by genetically modifying males to be sterile. Upon release, they outcompete their natural counterparts, reducing wild mosquito numbers, and in turn, lowering the risk of multiple diseases. In indoor cages, gene drives have fully suppressed a population of the insects within a year. Small-scale field tests are underway.

Gene drives have caught the eyes of plant scientists too, but initial efforts in plants failed.

The technology relies on CRISPR, which cuts DNA to insert, delete, or swap out genetic letters. Sensing damage to their DNA, cells activate internal molecular “repairmen” to stitch genes back together and adopt gene drives and their genetic cargo.

Plants are different. Their cells also have a DNA repair mechanism, but it’s only partially similar to that of insects or mice. Sticking a classic gene drive into plants can cause genetic mutations at the target site and even trigger resistance against the gene drive in a kind of a cellular civil war.

What Doesn’t Kill You Makes You Stronger

As a workaround, both new studies used a system dubbed “toxin-antidote.” Compared to previous gene drives, it doesn’t rely on canonical DNA repair.

The teams used a self-pollinating mustard plant for their studies. A darling in plant science research, its genome is well-known, and because the plant self-pollinates, it’s easier to contain the experiment. To build the gene drive, they developed a CRISPR-based method to destroy a gene that’s critical for survival called the “torpedo.” Any pollen without the gene can’t live on. A second construct, the “antidote,” carried a mimic of the same gene, but with modifications so that it’s resistant to destruction by CRISPR.

They examined two different genetic payloads. One study tinkered with a gene that’s essential to both male and female reproductive cells in plants. The other targeted a gene that disrupts pollen production.

Here’s the clever part: As the plant pollinates, offspring can inherit either the toxin, the antidote, or both. Only those with the antidote survive—plants that inherit the toxin rapidly die out. As a result, the system worked as a gene drive, with plants carrying the CRISPR-resistant gene taking over the population. The gene drives were highly efficient, passing down through generations roughly 99 percent of the time. And scientists didn’t see any signs of evolutionary adaptation—known as resistance—against the new genetic makeup.

Computer modeling showed the gene drive could overtake a single plant species in 10 to 30 generations. That’s impressive, according Neve and Barrett. Artificial genetic changes don’t often stick in wild plants—the plants tend to die off. The new gene drives suggest they could potentially last longer in the field, battling invasive species or cultivating hardier and pest-resistant crops that pass down beneficial traits over generations.

Despite their promise, gene drives remain controversial because of their potential to alter entire species. Scientists are still debating the ecological impacts. There’s also the concern that gene drives may hop over to unintended targets. For now, studies have designed genetic “brakes” to keep gene drives in check. Most studies are done in carefully controlled lab settings, and for malaria, potential unexpected consequences are being rigorously discussed before releasing gene drive-carrying mosquitos into the wild.

Even if the science works, the road to regulatory and societal approval may face roadblocks. Selling farmers on the technology may be difficult. And CRISPRed plants as a food source could also be tainted by the negative perception of genetically modified organisms (GMOs).

For now, the teams are looking towards a more acceptable everyday use—killing weeds. There are still a few kinks to work out. Gene drives only work when they can spread, so an ideal use is in plants that pollinate others, rather than those that self-pollinate, such as those in the studies. Still, the results are a proof of concept that the powerful technology can work in plants—though it may be awhile yet before it helps Henry with his knotweed problem.

Image Credit: Anthony Wade / Unsplash

Dune, widely considered one of the best sci-fi novels of all time, continues to influence how writers, artists, and inventors envision the future.

Of course, there are Denis Villeneuve’s visually stunning films, Dune: Part One (2021) and Dune: Part Two (2024).

But Frank Herbert’s masterpiece also helped Afrofuturist novelist Octavia Butler imagine a future of conflict amid environmental catastrophe; it inspired Elon Musk to build SpaceX and Tesla and push humanity toward the stars and a greener future; and it’s hard not to see parallels in George Lucas’ Star Wars franchise, especially the films’ fascination with desert planets and giant worms.

And yet when Herbert sat down in 1963 to start writing Dune, he wasn’t thinking about how to leave Earth behind. He was thinking about how to save it.

Herbert wanted to tell a story about the environmental crisis on our own planet, a world driven to the edge of ecological catastrophe. Technologies that had been inconceivable just 50 years prior had put the world at the edge of nuclear war and the environment on the brink of collapse; massive industries were sucking wealth from the ground and spewing toxic fumes into the sky.

When the book was published, these themes were front and center for readers, too. After all, they were living in the wake of both the Cuban missile crisis and the publication of Silent Spring, conservationist Rachel Carson’s landmark study of pollution and its threat to the environment and human health.

Dune soon became a beacon for the fledgling environmental movement and a rallying flag for the new science of ecology.

Indigenous Wisdoms

Though the term “ecology” had been coined almost a century earlier, the first textbook on ecology was not written until 1953, and the field was rarely mentioned in newspapers or magazines at the time. Few readers had heard of the emerging science, and even fewer knew what it suggested about the future of our planet.

While studying Dune for a book I’m writing on the history of ecology, I was surprised to learn that Herbert didn’t learn about ecology as a student or as a journalist.

Instead, he was inspired to explore ecology by the conservation practices of the tribes of the Pacific Northwest. He learned about them from two friends in particular.

The first was Wilbur Ternyik, a descendant of Chief Coboway, the Clatsop leader who welcomed explorers Meriwether Lewis and William Clark when their expedition reached the West Coast in 1805. The second, Howard Hansen, was an art teacher and oral historian of the Quileute tribe.

Ternyik, who was also an expert field ecologist, took Herbert on a tour of Oregon’s dunes in 1958. There, he explained his work to build massive dunes of sand using beach grasses and other deep-rooted plants in order to prevent the sands from blowing into the nearby town of Florence—a terraforming technology described at length in Dune.

As Ternyik explains he wrote for the US Department of Agriculture, his work in Oregon was part of an effort to heal landscapes scarred by European colonization, especially the large river jetties built by early settlers.

These structures disturbed coastal currents and created vast expanses of sand, turning stretches of the lush Pacific Northwest landscape into desert. This scenario is echoed in Dune, where the novel’s setting, the planet Arrakis, was similarly laid to waste by its first colonizers.

Hansen, who became the godfather to Herbert’s son, had closely studied the equally drastic impact logging had on the homelands of the Quileute people in coastal Washington. He encouraged Herbert to examine ecology carefully, giving him a copy of Paul B. Sears’ Where There Is Life, from which Herbert gathered one of his favorite quotes: “The highest function of science is to give us an understanding of consequences.”

The Fremen of Dune, who live in the deserts of Arrakis and carefully manage its ecosystem and wildlife, embody these teachings. In the fight to save their world, they expertly blend ecological science and Indigenous practices.

Treasures Hidden in the Sand

But the work that had the most profound impact on Dune was Leslie Reid’s 1962 ecological study The Sociology of Nature.

In it, Reid explained ecology and ecosystem science for a popular audience, illustrating the complex interdependence of all creatures within the environment.

“The more deeply ecology is studied,” Reid writes, “the clearer does it become that mutual dependence is a governing principle, that animals are bound to one another by unbreakable ties of dependence.”

In the pages of Reid’s book, Herbert found a model for the ecosystem of Arrakis in a surprising place: the guano islands of Peru. As Reid explains, the accumulated bird droppings found on these islands were an ideal fertilizer. Home to mountains of manure described as a new “white gold” and one of the most valuable substances on Earth, the guano islands became in the late 1800s ground zero for a series of resource wars between Spain and several of its former colonies, including Peru, Bolivia, Chile, and Ecuador.

At the heart of the plot of Dune is a battle for control of the “spice,” a priceless resource. Harvested from the sands of the desert planet, it’s both a luxurious flavoring for food and a hallucinogenic drug that allows some people to bend space, making interstellar travel possible.

There is some irony in the fact that Herbert cooked up the idea of spice from bird droppings. But he was fascinated by Reid’s careful account of the unique and efficient ecosystem that produced a valuable—albeit noxious—commodity.

As the ecologist explains, frigid currents in the Pacific Ocean push nutrients to the surface of nearby waters, helping photosynthetic plankton thrive. These support an astounding population of fish that feed hordes of birds, along with whales.

In early drafts of Dune, Herbert combined all of these stages into the life cycle of the giant sandworms, football-field-sized monsters that prowl the desert sands and devour everything in their path.

Herbert imagines each of these terrifying creatures beginning as small, photosynthetic plants that grow into larger “sand trout.” Eventually, they become immense sandworms that churn the desert sands, spewing spice onto the surface.

In both the book and Dune: Part One, soldier Gurney Halleck recites a cryptic verse that comments on this inversion of marine life and arid regimes of extraction: “For they shall suck of the abundance of the seas and of the treasure hid in the sand.”

‘Dune’ Revolutions

After Dune was published in 1965, the environmental movement eagerly embraced it.

Herbert spoke at Philadelphia’s first Earth Day in 1970, and in the first edition of the Whole Earth Catalog—a famous DIY manual and bulletin for environmental activists—Dune was advertised with the tagline: “The metaphor is ecology. The theme revolution.”

In the opening of Denis Villeneuve’s first adaptation of Dune, Chani, an indigenous Fremen played by Zendaya, asks a question that anticipates the violent conclusion of the second film: “Who will our next oppressors be?”

The immediate cut to a sleeping Paul Atreides, the white protagonist who’s played by Timothée Chalamet, drives the pointed anti-colonial message home like a knife. In fact, both of Villeneuve’s movies expertly elaborate upon the anti-colonial themes of Herbert’s novels.

Unfortunately, the edge of their environmental critique is blunted. But Villeneuve has suggested that he might also adapt Dune Messiah for his next film in the series—a novel in which the ecological damage to Arrakis is glaringly obvious.

I hope Herbert’s prescient ecological warning, which resonated so powerfully with readers back in the 1960s, will be unsheathed in Dune 3.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

We’ve all been there: A tight deadline, an overnighter, and the next day we’re navigating life like zombies.

For fighter pilots, the last step isn’t an option. During active duty, these pilots need to be in tip-top shape mentally, even when they’re deprived of sleep (which can be often). Typically, the treatment is your everyday cup of joe. But for longer durations of sleep deprivation, pilots are also prescribed stronger stimulants.

But as anyone who’s ever had too much caffeine knows, there are side effects. You get jittery. Your hands start to shake. Your mood takes a nosedive as the effect wears off and irritability sets in. And then you crash.

Prescription stimulants, such as dextroamphetamine, have even more severe side effects. As the name suggests, they’re in the same family as methamphetamine—or “meth”—and come with the risk of addiction. These drugs last longer inside the body, so that when trying to sleep after a tiring day, they keep parts of the brain in a semi-alert state and mess with sleep schedules. People taking dextroamphetamine often need sedatives to counteract lingering effects, and the chemical regime takes a toll.

Over time, the lack of restorative sleep impacts memory, cognition, and reasoning. It also damages the immune system, metabolism, and overall health.

The drugs work in short bursts. What if there’s a way to turn them on and off at will—giving the brain just a tiny dose when needed and quickly shutting off its effect to allow a full night’s sleep?

One solution may be light-activated drugs. The Defense Advanced Research Projects Agency (DARPA) announced a project in June to develop these types of drugs to combat sleep deprivation for fighter pilots. So-called photopharmacological drugs would add a molecular “light switch” to drugs like dextroamphetamine.

Pulses of light activate the drugs in parts of the brain on demand. Non-targeted brain regions aren’t exposed to the active version and continue to work normally. Once the pilots are alert, another pulse of light shuts off the drug, giving the body time to break it down before bedtime.

To make this vision a reality, the new project, Alert WARfighter Enablement (AWARE), has two research arms. One will develop safe and effective dextroamphetamine that can be controlled with light. The second will focus on engineering a wearable “helmet” of sorts to direct light pulses toward regions of the brain involved in alertness and mental acuity.

“To achieve the beneficial effects of stimulants on alertness without the undesirable effects of the stimulant on mood, restorative sleep, and mental health, a new approach is needed to enable targeted activation of the drug,” Dr. Pedro Irazoqui, AWARE program manager, said in a press release.

Brain on Alert

After a terrible night’s sleep, the first thing most of us reach for is coffee. Caffeine, its active ingredient, is the most widely used psychoactive substance in the world, with over 80 percent of people in North America drinking a cup of joe every morning.

While this is also the go-to solution for most fighter pilots, multiple countries have developed far stronger concoctions to keep their brigades awake. The most notorious is probably methamphetamine, first synthesized in the late 1800s. Best known by its street names—meth, crank, or speed—it was used during World War II to keep troops awake, before being outlawed across the globe. A safer spin-off, dextroamphetamine is currently prescribed to increase alertness and cognition. While effective, it can trigger both irritability and euphoric effects—a recipe for potential addiction.

The Air Force has approved other types of chemical drugs, such as modafinil, to battle fatigue too. Research in mice and people found these drugs can improve many cognitive functions—for example, navigating space, keeping multiple things in mind, and boosting overall alertness even when severely sleep-deprived. Unlike amphetamines, this group of drugs isn’t as addictive, with effects compared to drinking roughly 20 cups of coffee without the jitters. But they can produce pounding headaches, sweating, and in rare cases, hallucinations.

Light-activated drugs may be another option. First devised for cancer, these drugs have a molecular “light-switch” component that responds to pulses of light. The switch can be tagged onto conventional drugs, making it easy to adopt for existing medications—like, say, dextroamphetamine.

The “switch” component changes the chemical’s shape after being blasted by different wavelengths of light. Like transformers, one shape allows the chemical to grab onto its usual targets—the “active” state. Other configurations inactivate it.

Light-activated drugs have been tested in cells in petri dishes, but targeting the brain presents a hurdle—the skull. Shining a flashlight onto the skull obviously wouldn’t reach the brain, and invasive brain surgery is out of the question.

There’s a workaround. Infrared beams of light, at low levels, are safe in humans and can penetrate deep into tissues, including through the skull and into the brain. A previous study designed a number of potential switches that could be turned on with infrared light. And recent advances in AI could further aid the effort to develop “a photoswitchable version of dextroamphetamine that is inactive except in the presence of near-infrared light, which activates it,” wrote DARPA.

The other component is a programmable light-emitting helmet that transmits infrared light to the parts of the brain associated with wakefulness, reasoning, and decision-making. Over time, the stimulation could be personalized, so people only receive the necessary “dose” to stay alert.

The strategy still floods the brain with stimulants through a pill, but it limits the drug’s activity in time and space. With personalized dosages and light as a controller, it could lead to alertness without anxiety, irritability, or euphoria for each person. Switching the drug off also allows the brain to “rest” during a good night’s sleep.

A Three-Year Plan

AWARE is slated to last over three years. DARPA is now welcoming proposals that fit the program’s two goals, including developing light-activated dextroamphetamine, dubbed “PhotoDex,” that can be rapidly turned on and off in the presence of near-infrared light. All candidate drugs will first be validated in animal models, before moving on to human trials.

For the headset, the project envisions a setup that emits infrared light and reliably activates necessary parts of the brain at millimeter-resolution, roughly that of an MRI-based brain scan. The timeline is about a year, and the agency did not specify how the headsets should be designed—for example, wired or wireless, how they’re powered, or what mechanism turns on the light beams.

“The idea is very ambitious, but recent advances in the creation of phototherapeutics and light-emitting devices offer good reason to be optimistic about the prospects,” Dr. David Lawrence at the University of North Carolina, who is not involved in the project, told New Scientist.

For now, photoswitchable drugs have not yet been approved for human use. If the AWARE program goes as planned, it could open a new avenue for targeted drug treatment, not just for battling sleep deprivation, but also for other brain disorders. The project is well aware of the ethical, legal, and societal implications, and has plans to discuss the technology’s use.

Image Credit: US Air Force photo by 2nd Lt. Samuel Eckholm

 This is a recording of a 4-minute mini-keynote I did last month on Content vs. Context. I hope you enjoy and find it useful ???? Content is NOT King, Context Is! What does this mean? ???? Here it means ok, good or fine. But if you are in Brazil, it means you’re an ass. […]

Rats are incredibly nimble creatures. They can climb up curtains, jump down tall ledges, and scurry across complex terrain—say, your basement stacked with odd-shaped stuff—at mind-blowing speed.

Robots, in contrast, are anything but nimble. Despite recent advances in AI to guide their movements, robots remain stiff and clumsy, especially when navigating new environments.

To make robots more agile, why not control them with algorithms distilled from biological brains? Our movements are rooted in the physical world and based on experience—two components that let us easily explore different surroundings.

There’s one major obstacle. Despite decades of research, neuroscientists haven’t yet pinpointed how brain circuits control and coordinate movement. Most studies have correlated neural activity with measurable motor responses—say, a twitch of a hand or the speed of lifting a leg. In other words, we know brain activation patterns that can describe a movement. But which neural circuits cause those movements in the first place?

We may find the answer by trying to recreate them in digital form. As the famous physicist Richard Feynman once said, “What I cannot create, I do not understand.”

This month, Google DeepMind and Harvard University built a realistic virtual rat to home in on the neural circuits that control complex movement. The rat’s digital brain, composed of artificial neural networks, was trained on tens of hours of neural recordings from actual rats running around in an open arena.

Comparing activation patterns of the artificial brain to signals from living, breathing animals, the team found the digital brain could predict the neural activation patterns of real rats and produce the same behavior—for example, running or rearing up on hind legs.

The collaboration was “fantastic,” said study author Dr. Bence Ölveczky at Harvard in a press release. “DeepMind had developed a pipeline to train biomechanical agents to move around complex environments. We simply didn’t have the resources to run simulations like those, to train these networks.”

The virtual rat’s brain recapitulated two regions especially important for movement. Tweaking connections in those areas changed motor responses across a variety of behaviors, suggesting these neural signals are involved in walking, running, climbing, and other movements.

“Virtual animals trained to behave like their real counterparts could provide a platform for virtual neuroscience…that would otherwise be difficult or impossible to experimentally deduce,” the team wrote in their article.

A Dense Dataset

Artificial intelligence “lives” in the digital world. To power robots, it needs to understand the physical world.

One way to teach it about the world is to record neural signals from rodents and use the recordings to engineer algorithms that can control biomechanically realistic models replicating natural behaviors. The goal is to distill the brain’s computations into algorithms that can pilot robots and also give neuroscientists a deeper understanding of the brain’s workings.

So far, the strategy has been successfully used to decipher the brain’s computations for vision, smell, navigation, and recognizing faces, the authors explained in their paper. However, modeling movement has been a challenge. Individuals move differently, and noise from brain recordings can easily mess up the resulting AI’s precision.

This study tackled the challenges head on with a cornucopia of data.

The team first placed multiple rats into a six-camera arena to capture their movement—running around, rearing up, or spinning in circles. Rats can be lazy bums. To encourage them to move, the team dangled Cheerios across the arena.

As the rats explored the arena, the team recorded 607 hours of video and also neural activity with a 128-channel array of electrodes implanted in their brains.

They used this data to train an artificial neural network—a virtual rat’s “brain”—to control body movement. To do this, they first tracked how 23 joints moved in the videos and transferred them to a simulation of the rats’ skeletal movements. Our joints only bend in certain ways, and this step filters out what’s physically impossible (say, bending legs in the opposite direction).

The core of the virtual rat’s brain is a type of AI algorithm called an inverse dynamics model. Basically, it knows where “body” positions are in space at any given time and, from there, predicts the next movements leading to a goal—say, grab that coffee cup without dropping it.

Through trial-and-error, the AI eventually came close to matching the movements of its biological counterparts. Surprisingly, the virtual rat could also easily generalize motor skills to unfamiliar places and scenarios—in part by learning the forces needed to navigate the new environments.

The similarities allowed the team to compare real rats to their digital doppelgangers, when performing the same behavior.

In one test, the team analyzed activity in two brain regions known to guide motor skills. Compared to an older computational model used to decode brain networks, the AI could better simulate neural signals in the virtual rat across multiple physical tasks.

Because of this, the virtual rat offers a way to study movement digitally.

One long-standing question, for example, is how the brain and nerves command muscle movement depending on the task. Grabbing a cup of coffee in the morning, for example, requires a steady hand without any jerking action but enough strength to hold it steady.

The team tweaked the “neural connections” in the virtual rodent to see how changes in brain networks alter the final behavior—getting that cup of coffee. They found one network measure that could identify a behavior at any given time and guide it through.

Compared to lab studies, these insights “can only be directly accessed through simulation,” wrote the team.

The virtual rat bridges AI and neuroscience. The AI models here recreate the physicality and neural signals of living creatures, making them invaluable for probing brain functions. In this study, one aspect of the virtual rat’s motor skills relied on two brain regions—pinpointing them as potential regions key to guiding complex, adaptable movement.

A similar strategy could provide more insight into the computations underlying vision, sensation, or perhaps even higher cognitive functions such as reasoning. But the virtual rat brain isn’t a complete replication of a real one. It only captures snapshots of part of the brain. But it does let neuroscientists “zoom in” on their favorite brain region and test hypotheses quickly and easily compared to traditional lab experiments, which often take weeks to months.

On the robotics side, the method adds a physicality to AI.

“We’ve learned a huge amount from the challenge of building embodied agents: AI systems that not only have to think intelligently, but also have to translate that thinking into physical action in a complex environment,” said study author Dr. Matthew Botvinick at DeepMind in a press release. “It seemed plausible that taking this same approach in a neuroscience context might be useful for providing insights in both behavior and brain function.”

The team is next planning to test the virtual rat with more complex tasks, alongside its biological counterparts, to further peek inside the inner workings of the digital brain.

“From our experiments, we have a lot of ideas about how such tasks are solved,” said Ölveczky to The Harvard Gazette. “We want to start using the virtual rats to test these ideas and help advance our understanding of how real brains generate complex behavior.”

Image Credit: Google DeepMind

Imagine a future with nearly silent air taxis flying above traffic jams and navigating between skyscrapers and suburban droneports. Transportation arrives at the touch of your smartphone and with minimal environmental impact.

This isn’t just science fiction. United Airlines has plans for these futuristic electric air taxis in Chicago and New York. The US military is already experimenting with them. And one company has a contract to launch an air taxi service in Dubai as early as 2025. Another company hopes to defy expectations and fly participants at the 2024 Paris Olympics.

Backed by billions of dollars in venture capital and established aerospace giants that include Boeing and Airbus, startups across the world such as Joby, Archer, Wisk, and Lilium are spearheading this technological revolution, developing electric vertical takeoff and landing (eVTOL) aircraft that could transform the way we travel.

Electric aviation promises to alleviate urban congestion, open up rural areas to emergency deliveries, slash carbon emissions, and offer a quieter, more accessible form of short-distance air travel.

But the quest to make these electric aircraft ubiquitous across the globe instead of just playthings for the rich is far from a given. Following the industry as executive director of the Oklahoma Aerospace Institute for Research and Education provides a view of the state of the industry. Like all great promised paradigm shifts, numerous challenges loom—technical hurdles, regulatory mazes, the crucial battle for public acceptance, and perhaps physics itself.

Why Electrify Aviation?

Fixed somewhere between George Jetson’s flying car and the gritty taxi from The Fifth Element, the allure of electric aviation extends beyond gee-whiz novelty. It is rooted in its potential to offer efficient, eco-friendly alternatives to ground transportation, particularly in congested cities or hard-to-reach rural regions.

While small electric planes are already flying in a few countries, eVTOLs are designed for shorter hops—the kind a helicopter might make today, only more cheaply and with less impact on the environment. The eVTOL maker Joby purchased Uber Air to someday pair the company’s air taxis with Uber’s ride-hailing technology.

In the near term, once eVTOLs are certified to fly as commercial operations, they are likely to serve specific, high-demand routes that bypass road traffic. An example is United Airlines’ plan to test Archer’s eVTOLs on short hops from Chicago to O’Hare International Airport and Manhattan to Newark Liberty International Airport.

While some applications initially might be restricted to military or emergency use, the goal of the industry is widespread civil adoption, marking a significant step toward a future of cleaner urban mobility.

The Challenge of Battery Physics

One of the most significant technical challenges facing electric air taxis is the limitations of current battery technology.

Today’s batteries have made significant advances in the past decade, but they don’t match the energy density of traditional hydrocarbon fuels currently used in aircraft. This shortcoming means that electric air taxis cannot yet achieve the same range as their fossil-fueled counterparts, limiting their operational scope and viability for long-haul flights. Current capabilities still fall short of traditional transportation. However, with ranges from dozens of miles to over 100 miles, eVTOL batteries provide sufficient range for intracity hops.

The quest for batteries that offer higher energy densities, faster charging times, and longer life cycles is central to unlocking the full potential of electric aviation.

While researchers are working to close this gap, hydrogen presents a promising alternative, boasting a higher energy density and emitting only water vapor. However, hydrogen’s potential is tempered by significant hurdles related to safe storage and infrastructure capable of supporting hydrogen-fueled aviation. That presents a complex and expensive logistics challenge.

And, of course, there’s the specter of the last major hydrogen-powered aircraft. The Hindenburg airship caught fire in 1937, but it still looms large in the minds of many Americans.

Regulatory Hurdles

Establishing a “4D highways in the sky” will require comprehensive rules that encompass everything from vehicle safety to air traffic management. For the time being, the US Federal Aviation Administration is requiring that air taxis include pilots serving in a traditional role. This underscores the transitional phase of integrating these vehicles into airspace, highlighting the gap between current capabilities and the vision of fully autonomous flights.

The journey toward autonomous urban air travel is fraught with more complexities, including the establishment of standards for vehicle operation, pilot certification, and air traffic control. While eVTOLs have flown hundreds of test flights, there have also been safety concerns after prominent crashes involving propeller blades failing on one in 2022 and the crash of another in 2023. Both were being flown remotely at the time.

The question of who will manage these new airways remains an open discussion—national aviation authorities such as the FAA, state agencies, local municipalities, or some combination thereof.

Creating the Future

In the long term, the vision for electric air taxis aligns with a future where autonomous vehicles ply the urban skies, akin to scenes from Back to the Future. This future, however, not only requires technological leaps in automation and battery efficiency but also a societal shift in how people perceive and accept the role of autonomous vehicles, both cars and aircraft, in their daily lives. Safety is still an issue with autonomous vehicles on the ground.

The successful integration of electric air taxis into urban and rural environments hinges on their ability to offer safe, reliable, and cost-effective transportation.

As these vehicles overcome the industry’s many hurdles, and regulations evolve to support their operation in the years ahead, I believe we could witness a profound transformation in air mobility. The skies offer a new layer of connectivity, reshaping cities and how we navigate them.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Image Credit: Joby

Thousands of people a year die while waiting for an organ transplant. Early experiments in xenotransplantation are raising hopes this could soon be a thing of the past.

In the US, 100,000 people are currently on the organ transplant waiting list, and 17 of them die every day before receiving an organ. The persistent shortage of organ donors has long led doctors to flirt with the idea of xenotransplantation, a procedure where tissue or an organ from an animal is transplanted into a human.

Early experiments were largely unsuccessful and ethically questionable, though, and the idea remained firmly on the fringes of the medical world. That’s largely due to the high risk of rejection. This is a problem for human transplants too, but it’s much more risky when using organs from other species.

But the advent of increasingly powerful and precise genetic engineering technologies such as CRISPR have ushered the idea from the shadows. The ability to make edits to the donor animals’ DNA to prevent the production of biomolecules known to induce immune responses in humans has raised hopes the approach may be viable after all.

In recent years, a handful of pioneering experiments in humans have demonstrated that genetically engineered pig organs can at least temporarily function smoothly in the human body. Medical complications, organ rejections, and patient deaths have meant none of these procedures have provided a long-term solution, but the results so far have been promising.

“At Massachusetts General Hospital alone, there are over 1,400 patients on the waiting list for a kidney transplant,” Leonardo Riella, who led the surgical team at Mass General that transplanted a pig kidney into a patient, said in a press release earlier this year.

“Some of these patients will unfortunately die or get too sick to be transplanted due to the long waiting time on dialysis. I am firmly convinced that xenotransplantation represents a promising solution to the organ shortage crisis.”

In 2021, in the first human experiment involving a genetically engineered pig organ, doctors transplanted a kidney into a patient who was already brain dead. The team knocked out a gene for a molecule called alpha-gal—which causes organ rejection—in the donor pig. The surgery appeared to be a success: The kidney produced urine and showed no signs of rejection, but the patient was only kept alive for 54 hours.

The following year, a patient with terminal heart failure received a genetically modified pig heart and initially seemed to do well, but then passed away 60 days later. While it’s not entirely clear why he died, the doctors found that pre-screening failed to flag a pathogen called porcine cytomegalovirus that was found in his heart afterwards, which could have contributed. He’d also been given an antibody treatment that had reacted with the heart.

Then earlier this year, two kidney disease patients who were ineligible for normal transplants received gene-edited pig kidneys from donor pigs bred by biotech firm eGenesis. Using CRISPR, the company made 69 edits that removed some pig genes, added some human ones, and reduced the risk of latent virus in the organ reactivating and harming the patient.

The procedures appeared to go well. Doctors even discharged the first patient after determining the kidney was functioning well, and he no longer needed dialysis. Two months later he passed away, but he had other underlying health issues, and the hospital said there was no indication his death was the result of the transplant.

The second patient had to have the kidney removed after 47 days due to “unique challenges” stemming from the fact she had also had a mechanical heart pump implanted just before the transplantation. There were no signs of rejection, but the kidney started losing function because her heart was not able to pump blood with enough pressure, the researchers said.

The most recent experiment was announced in May, when Chinese researchers said they had transplanted a liver from a genetically modified pig into a 71-year-old man with liver cancer. While details of the procedure are limited, the team claimed the man was “doing very well” more than two weeks after surgery.

While most of these experiments have been short-lived, the fact that only two cases saw the transplanted organ fail—one of which was due to external complications—is a promising sign. For ethical reasons, doctors have only been able to experiment with patients whose chances of survival were already slim.

But it does mean that we have little idea whether xenotransplantation could be a viable long-term solution for patients. There is also some concern that implanting organs from other animals into humans could make it easier for pathogens to jump between species, potentially creating the risk of new pandemics.

Other researchers are investigating whether, instead of transplanting pig organs into humans, we could grow human organs in pigs. Last September, researchers announced they’d transplanted human stem cells into pig embryos where they then grew into rudimentary kidneys.

This approach is a long way from human trials though, so for the time being, xenotransplantation seems like a more promising way to bring down transplant wait times. While it’s still early days, the promising early results suggest we may not be far from a future where replacement organs can be grown to order.

Image Credit: Massachusetts General Hospital

If you had to sum up what has made humans such a successful species, it’s teamwork. There’s growing evidence that getting AIs to work together could dramatically improve their capabilities too.

Despite the impressive performance of large language models, companies are still scrabbling for ways to put them to good use. Big tech companies are building AI smarts into a wide-range of products, but none has yet found the killer application that will spur widespread adoption.

One promising use case garnering attention is the creation of AI agents to carry out tasks autonomously. The main problem is that LLMs remain error-prone, which makes it hard to trust them with complex, multi-step tasks.

But as with humans, it seems two heads are better than one. A growing body of research into “multi-agent systems” shows that getting chatbots to team up can help solve many of the technology’s weaknesses and allow them to tackle tasks out of reach for individual AIs.

The field got a significant boost last October when Microsoft researchers launched a new software library called AutoGen designed to simplify the process of building LLM teams. The package provides all the necessary tools to spin up multiple instances of LLM-powered agents and allow them to communicate with each other by way of natural language.

Since then, researchers have carried out a host of promising demonstrations. 

In a recent article, Wired highlighted several papers presented at a workshop at the International Conference on Learning Representations (ICLR) last month. The research showed that getting agents to collaborate could boost performance on math tasks—something LLMs tend to struggle with—or boost their reasoning and factual accuracy.

In another instance, noted by The Economist, three LLM-powered agents were set the task of defusing bombs in a series of virtual rooms. The AI team performed better than individual agents, and one of the agents even assumed a leadership role, ordering the other two around in a way that improved team efficiency.

Chi Wang, the Microsoft researcher leading the AutoGen project, told The Economist that the approach takes advantage of the fact most jobs can be split up into smaller tasks. Teams of LLMs can tackle these in parallel rather than churning through them sequentially, as an individual AI would have to do.

So far, setting up multi-agent teams has been a complicated process only really accessible to AI researchers. But earlier this month, the Microsoft team released a new “low-code” interface for building AI teams called AutoGen Studio, which is accessible to non-experts.

The platform allows users to choose from a selection of preset AI agents with different characteristics. Alternatively, they can create their own by selecting which LLM powers the agent, giving it “skills” such as the ability to fetch information from other applications, and even writing short prompts that tell the agent how to behave. 

So far, users of the platform have put AI teams to work on tasks like travel planning, market research, data extraction, and video generation, say the researchers.

The approach does have its limitations though. LLMs are expensive to run, so leaving several of them to natter away to each other for long stretches can quickly become unsustainable. And it’s unclear whether groups of AIs will be more robust to mistakes, or whether they could lead to cascading errors through the entire team.

Lots of work needs to be done on more prosaic challenges too, such as the best way to structure AI teams and how to distribute responsibilities between their members. There’s also the question of how to integrate these AI teams with existing human teams. Still,  pooling AI resources is a promising idea that’s quickly picking up steam.

Image Credit: Mohamed Nohassi / Unsplash