How Quickly Do Large Language Models Learn Unexpected Skills?

Two years ago, in a project called the Beyond the Imitation Game benchmark, or BIG-bench, 450 researchers compiled a list of 204 tasks designed to test the capabilities of large language models, which power chatbots like ChatGPT. On most tasks, performance improved predictably and smoothly as the models scaled up — the larger the model,… Continue reading How Quickly Do Large Language Models Learn Unexpected Skills?

A ‘Lobby’ Where a Molecule Mob Tells Genes What to Do

That’s why well-controlled experiments like those of the Stanford team are so useful, Kuhlman added. Usually, when researchers need to measure weak interactions like these, they have two choices: They can make a few very detailed, extremely precise measurements and generalize from them, or they can take a great many quick-and-dirty measurements and use mathematically… Continue reading A ‘Lobby’ Where a Molecule Mob Tells Genes What to Do

AI Audio Deepfakes Are Quickly Outpacing Detection

Experts have long warned of a future where artificial intelligence makes it impossible to tell digital fact from fiction. Now that future is here. A recent case of a recording that sounds like a high school principal making racist comments shows the risk that widely available generative AI tools can pose and the difficulty of… Continue reading AI Audio Deepfakes Are Quickly Outpacing Detection

Cleaning Water Naturally the Ancient Maya Way

Water is life. That’s why we need to take care of it. Even plentiful water supplies are moot if they are undrinkable. Climate change, pollution and growing populations only add to the urgency of maintaining adequate water supplies and water quality for humanity. After doing archaeology for 35 years in Belize, focusing on the ancestral… Continue reading Cleaning Water Naturally the Ancient Maya Way

AI Survey Exaggerates Apocalyptic Risks

The headlines in early January didn’t mince words, and all were variations on one theme: researchers think there’s a 5 percent chance artificial intelligence could wipe out humanity. That was the sobering finding of  a paper posted on the preprint server arXiv.org. In it, the authors reported the results of a survey of 2,778 researchers… Continue reading AI Survey Exaggerates Apocalyptic Risks

Researchers Approach New Speed Limit for Seminal Problem

The traveling salesperson problem is one of the oldest known computational questions. It asks for the ideal route through a certain list of cities, minimizing mileage. Despite seeming simple, the problem is notoriously difficult. While you can use brute force to check all the possible routes until you find the shortest path, such a strategy… Continue reading Researchers Approach New Speed Limit for Seminal Problem

Consciousness Is a Continuum, and Scientists Are Starting to Measure It

What does it mean to be conscious? People have been thinking and writing about this question for millennia. Yet many things about the conscious mind remain a mystery, including how to measure and assess it. What is a unit of consciousness? Are there different levels of consciousness? What happens to consciousness during sleep, coma and… Continue reading Consciousness Is a Continuum, and Scientists Are Starting to Measure It

Hospice Providers Must Be Better Regulated

February 1, 2024 4 min read Too many hospice providers in the U.S. are run by private equity and for-profit corporations. A lack of regulation allows them to provide abysmal end-of-life care By The Editors Everyone deserves a good death—a choice about how they spend their final days, a peaceful, pain-free exit. That is the… Continue reading Hospice Providers Must Be Better Regulated