AI scientists

Attention is turning to Elizabeth Feinler, the information scientist who led the Network Information Center for ARPANET, the precursor to the modern internet. Her group managed hostnames and directories that laid early foundations for how the internet is organised today. Discussions tie her legacy to current debates on cybersecurity, online investigations and digital history, with renewed interest in the pioneers behind the internet's early infrastructure.

A new artificial intelligence tool is being presented as a way to help scientists draw connections across separate fields, enabling the kind of rare cross-disciplinary breakthroughs that researchers rarely make on their own. The claim is that AI can spot links between bodies of knowledge that human specialists, working within their own disciplines, tend to miss.

A newly reported AI tool is designed to help scientists make cross-disciplinary breakthroughs that have historically been rare. By surfacing unexpected connections between fields, the tool could make 'borrowed' insights — like how physics ideas find uses in biology or medicine — a more routine part of research. Commenters see promise in accelerating discovery, though broader details on performance and adoption remain limited in early coverage.

Researchers have used artificial intelligence to design proteins that act as proteases, enzymes capable of cutting up other proteins. The work, reported by Ars Technica, marks a step beyond AI-designed binding proteins toward custom enzymes with real catalytic function. Scientists say such designer enzymes could eventually be used in medicine, research and industry, though the full details and applications remain to be seen.

Scientists have issued a warning cataloguing 17 major threats to humanity, stressing that artificial intelligence and nuclear war are not the only existential dangers. The list is said to cover a broad range of risks, from environmental and biological hazards to technological ones. The warning is drawing attention for expanding the conversation on global catastrophic risks beyond the two most commonly cited threats.

Mathematician Terence Tao has published a new essay asking "Why are we even doing all this?", examining the purpose of mathematical work now that large language models can assist with research. The piece is being shared widely, including among non-scientists, with readers noting it articulates conversations already happening in STEM circles about how AI is changing the value and meaning of intellectual labour.

Scientists are raising concerns about the environmental impact of artificial intelligence, focusing on the ecological cost of the data centers that power AI systems. The facilities consume large amounts of electricity and water, and researchers are calling for greater scrutiny of their footprint as AI use expands worldwide.

Commenters are drawing on the Frankenstein myth to describe the ambition behind artificial intelligence: the age-old dream of breathing a Promethean spark into inanimate matter and recapitulating the creation of life. The post reflects a wider debate about whether modern AI research amounts to a Promethean act, with opinions split between fascination with the goal and caution about scientists' hubris in attempting it.

MIT researchers have developed an artificial intelligence model aimed at improving research in material science. The model is expected to help scientists analyse and predict material properties more efficiently, potentially accelerating discoveries in fields such as energy and manufacturing. Details on the model's capabilities and results have not yet been widely reported, but the announcement has drawn attention across science news outlets.

Scientists are using artificial intelligence to identify the best conditions and times to hunt Burmese pythons in Florida. The models aim to improve detection of the invasive snakes, which are notoriously hard to find and have devastated wildlife in the Everglades. The approach could make removal efforts by hunters and wildlife agencies more effective.