Nicola Jones

As you may already know, I write the Nature Briefing: AI and Robotics , a biweekly newsletter sent straight to email inboxes with the latest and greatest news in AI as it pertains to science. You should sign up, if you haven't already. Part of my job is to check the inbox to see if / how people have responded to this newsletter. Most of what I get is bounce-backs from peoples' out-of-office alerts, or spam, along with the occasional thoughtful reply, which I respond to. Recently I noticed that the longest, most thoughtful replies were coming from 'Hector', who was really reading my emails with the most careful attention and had interesting things to say in response. That said, the replies were coming suspiciously quickly -- I mean seconds after the original emails went out. This seemed... um... unlikely. Upon googling the email address (coze.email) I discovered that 'coze' is indeed an AI assistant service, which uses agents to reply to emails. I presume this is...

The world's most-famous preprint server, arXiv, which hosts millions of papers in physics, computer science and more, is leaving the nest of its university origins to go out on its own this year. Read all about it in my Q&A for the Foundational Questions Institute .

A simple strategy of hunting for odd language errors might help finger fake research papers churned out by so-called paper mills, forensic metascientist James Heathers of the Medical Evidence Project reported here at the World Conference on Research Integrity last week. By automating the search for such errors, science sleuths could potentially identify large numbers of problematic papers.  Read my story in Science . 

These days, thanks to AI, pretty much anyone can code -- or make the apps, websites and data processing pipelines that used to require a knowledge of code. Don't know how to do it? Just ask your friendly neighbourhood AI to do it for you. Pretty much all professional developers are now vibe coding to some degree, and researchers are doing it too. Most are seriously impressed with how this speeds up their work and frees up their creativity. But it comes with a lot of potential pitfalls and caveats.  Read my story on vibe coding in Nature . 

Researchers, publishers and ethicists are grappling with when, how and why to disclose the use of AI in scientific work. My story for Science from the World Conference on Research Integrity. 

Artificial intelligence (AI) tools designed to execute end-to-end projects, from coming up with hypotheses to running and writing up experiments, are increasingly popular with researchers—and increasingly skilled. But a new study shows these tools can stealthily violate norms of research integrity. Read my story for Science from the World Conference on Research Integrity. And... Nihar must win some kind of award for putting this much effort into his presentation: https://www.dropbox.com/scl/fi/u3nj3fzknpjl9rsnp5cxx/grim_reaper_intro.m4v?rlkey=7ow42i038wovzelj8rxt2vec4&e=1&st=5cii2vwh&dl=0 

As the world careens past our hoped-for target of 1.5 degrees Celsius warming, scientists are growing increasingly alarmed that we may be nearing a dramatic, long-feared “tipping point” — a moment when the main ocean current in the Atlantic Ocean becomes destined to shut down, clamping off the primary source of warmth for northern Europe and playing havoc with the global climate. Such a scenario has been a concern for many decades, but the issue is now heating up. “I have personally researched this for 35 years,” says Stefan Rahmstorf, a physical oceanographer at the Potsdam Institute for Climate Impact Research in Germany. “For the first 30 years we considered this a low likelihood event — I would have said a 5 percent chance of occurring. It’s more like 50/50 now. I would even say more likely than not.” Read my feature in Yale Environment 360