Andrew L. Hufton

@alhufton.bsky.social

I'm just getting started on Bluesky... Personal website: https://alhufton.com/. Editor-in-Chief of Patterns at Cell Press @cp-patterns.bsky.social

The things that I find most interesting about this paper are the differences between the tested models. The OpenAI and Claude models tend to perform similarly and have a common weakness in "no-evidence" contexts, i.e. they like to make up an answer even if there isn't medical evidence.

Patterns, a Cell Press journal@cp-patterns.bsky.social · 4mo ago

Online Now: Evaluating large language models for evidence-based clinical question answering #datascience

If you’re going to write a tutorial paper describing a tool or method, it makes a world of difference showing the paper to someone unfamiliar with the tool and asking them to recreate the examples. Some of these papers can be really tricky to follow

Unbelievable. This would be a terrible blow to American science, writ large. It would decimate not only climate research, but also the kind of weather, wildfire, and disaster research that has underpinned half a century of progress in prediction, early warning, and increased resilience.

USA TODAY@usatoday.com · 8mo ago

Exclusive: The Trump administration is moving to dismantle the National Center for Atmospheric Research in Colorado, according to a senior White House official, taking aim at one of the world's leading climate research labs.

2️⃣ Researchers tested this idea using a large-scale controlled resume correspondence experiment. SCARY Result: LLMs consistently preferred resumes written by themselves over human or rival model outputs — even when quality was the same. 2/n

Sundar Balakrishnan@drsundar.bsky.social · 9mo ago

Thread: When AIs Hire Themselves 🧵 1️⃣ LLMs are now in play on both sides of hiring — candidates use them to polish resumes, while employers use them to screen applicants. A new study asks: do LLMs favor their own outputs? 1/n

1/2 Bacteriophages are among Earth’s most diverse biological entities, but many resist genetic manipulation. This leaves much of their diversity unexplored and constrains efforts to discover new proteins and exploit their potential for biotechnology and therapeutic applications.

Are you an educator in a higher ed setting using generative AI in innovative ways in the classroom? We are currently seeking contributors for a short piece being developed at the journal on the topic. Please DM us if you would be interested in sharing your story.