iiinova — where your digital self keeps evolving

This paper is a first look at how AlphaFold is impacting science, but this story will continue as AI tools develop and the research community adapts to new opportunities. There's a lot to learn here, so we would love to talk to anyone who is working on similar topics!

000

These results suggest that AI is facilitating new research, but progress is slow. A bottlenecks model of R&D production seems to be at play, where AI unlocks tasks, lowering the cost of new discovery, but downstream progress is slowed by tasks that remain difficult to automate.

100

Lastly, we study impacts on early stage drug discovery by looking at bioactivity experiments, a key first step in in assessing whether candidate drugs bind to target proteins. We see no evidence yet of a shift in attention toward previously-undetermined proteins.

100

We find a noticeable shift in attention toward previously unsolved structures that now have new structural insights available through AlphaFold. These papers are about a variety of topics, including protein function, process, and diseases.

100

Next we turn our attention to the broader field of protein science and consider how scientists in related areas are using AlphaFold insights. Here we compare publishing activity about proteins that had previously been experimentally solved vs. those that had not

100

Instead of AI substituting for experimental output, we see evidence of complementarities between experimental and AI methods. They are now using AlphaFold structures as templates to help solve new structures experimentally.

100

We see no evidence of this so far. Here we plot the number of new experimentally-determined structures in the PDB in the three years since AlphaFold. They continue to publish these structures and do so in good journals like Science, Nature, and Cell.

100

The first part of our empirical analysis focuses on AlphaFold as a task-automating technology. We ask: Has the introduction of AlphaFold started to displace experimentalists, whose (rather expensive) methods might be substitutable with AI output?

100

Between 2021 and 2022, DeepMind predicted the structure of every known protein and posted the structures publicly on an EMBL-hosted website. A huge trove of new structure insights immediately became available for public use. Exciting stuff!

100

In 2020, DeepMind introduced AlphaFold2 in the CASP competition, and provided the first AI structure predictions with near-experimental accuracy. It was heralded as a huge and unexpected leap forward, with many predicting a rapid acceleration in protein science and drug discovery

100

Proteins can be described as a 2d sequence of amino acids that fold together to make a 3d structure. It has long been hypothesized that computational methods could be used to predict 3d structures of proteins from just the sequence.

100

For decades, structural biologists have used expensive and difficult experimental methods like x-ray crystallography to solve protein structures. Even after half a century of effort, there are many proteins that have not been solved experimentally.

100

A quick preview of the results: 1. AlphaFold is complementing rather than substituting for experimental research 2. AlphaFold is stimulating new basic research about previously unsolved proteins 3. No evidence yet that it has had a noticeable impact on early-stage drug research

100

The paper is preliminary and our analysis is ongoing, but due to the excitement around the topic, we wanted to report what we've learned so far. Feedback is welcome! Paper link: carolynstein.github.io/files/papers...

carolynstein.github.io

100

mammon @derekhill.bsky.social · 06/11/2025

There is great excitement about the potential for AI to reshape science, but so far very little empirical evidence about how that is (or is not??) happening in real time. I'm excited to share a new working paper with Carolyn Stein about the impact of AlphaFold on science ->

176

Reposted by Ryan Hill

i'm never showing my kid tony hawk. they'll have to recreate the 900 from first principles

762251

Ryan Hill @ryanreedhill.bsky.social · 29/09/2025

*mulder’s eyes widen as scully hands him the ancient xylophone*

010

Ryan Hill @ryanreedhill.bsky.social · 14/08/2025

You listed six different publishers that are operating at scale. That’s more competitive than many mature industries, and it doesn’t include the many non-profit and university publishers

000

Ryan Hill @ryanreedhill.bsky.social · 13/08/2025

The best way to think about academic publishing is that the creators are also the consumers. No one is “buying” our papers at the newsstand. Elsevier is basically a management company that gets a fee for running the peer-review process. If they charge too much, a competitor will replace them.

150

Ryan Hill @ryanreedhill.bsky.social · 11/08/2025

Henry Hudson to blame presumably

020

Ryan Hill @ryanreedhill.bsky.social · 07/08/2025

Love that 5% of people don’t care for maps on their phones. Give me the Rand McNally US Road Atlas or give me death!

030

Ryan Hill @ryanreedhill.bsky.social · 04/08/2025

He can ask the cash cab questions

Ryan Hill @ryanreedhill.bsky.social · 08/06/2025

If we’re being honest, straw is the best roofing material of the three. The pig appreciates the risks of damp

120

Ryan Hill @ryanreedhill.bsky.social · 28/05/2025

We didn’t look at that specifically, but it’s an interesting question for further research! My hunch is that experience is still a benefit even when pivoting. Other work on the “Matthew effect” suggests it’s hard to break into science at all. Thanks for the comment!

Ryan Hill @ryanreedhill.bsky.social · 28/05/2025

Thanks to great co-authors @yianyin.bsky.social, Carolyn Stein, Xizhao Wang, @dashunwang.bsky.social, and Ben Jones. It was a long time coming!

Ryan Hill @ryanreedhill.bsky.social · 28/05/2025

Very excited for our new publication in @nature.com about the difficulty of pivoting in science. In some sense, this project was my own "pivot" to engage more with a general science audience for the first time. Its really gratifying to see it in print.

272

Ryan Hill @ryanreedhill.bsky.social · 17/05/2025

Dandelion infestations in suburban yards as a leading indicator for recession

020

The whisper network remains influential and necessary

160

This case in particular is a reminder that "proprietary data" and "anonymous partners" can occasionally be permitted for social science research, but should raise additional layers of scrutiny in the peer review process

1113

MIT Says It No Longer Stands Behind Student’s AI Research Paper

With the excitement (delirium?) surrounding the rise of AI, it's more important than ever that scientists and journalists maintain an attitude of skepticism and dispassion about new findings as they arrive www.wsj.com/tech/ai/mit-...

wsj.com

The university said it has no confidence in a widely circulated paper by an economics graduate student.

43912

Ryan Hill @ryanreedhill.bsky.social · 27/04/2025

It just all seemed way too good to be true. I've had a number of skeptical conversations about this paper with other folks working in the area. We hesitated to publicly question the paper because he's a grad student and it wasn't peer reviewed yet. Glad the proper channels were followed to fix it.

1120

A family friend went to Salzburg but didn’t want to pay for the TSOM bus tour. So instead they followed the bus and tried to figure out the sites on their own. After an hour they realized they accidentally were following the history of the Nazi occupation tour instead