Every system that was regulated, either explicitly or implicitly, by the fact that they were effortful for humans (letters of recommendation, government filings, essays, or, as this paper finds, lawsuits) will break under a wave of AI.
Their mental load was “reduced” by having a computer electrically stimulate their arm instead.
Bodily autonomy wise, it might feel a bit freaky, because you have the proprioception of your arm moving, but without the mental load of you moving it.
I wish more research was poured in this area.
Would you let a computer hijack your muscle movements if it increased your performance 35%?
I totally would.
Came across a really interesting ACM paper today (SplitBody), where subjects were given difficult multitasking challenges.
If you are making a recommendation on how to improve Ai that goes something like "design ai to admit uncertainty/when it's wrong/challenge the user" you need to specify how you expect that happen. Sycophancy is a well known issue. How will you change model maker incentives? Regulate? Something else?
I'm co-organizing a free, 2-day Zoom workshop from the NAS & hope you'll consider attending!
📅 April 23-24: Discuss how we can enhance scientific integrity in the social and behavioral sciences
REGISTER: www.nationalacademies.org/projects/DBA...
PROGRAM: www.nationalacademies.org/cdn/material...
CodaNote: Collaborative Markdown/Quarto/Rmarkdown
People were looking for collaborating on MD family files with knit rendering...
I built this tool for CORE team, works like Google Docs.
Opened it to all from internal use so everyone can use it.
Feedback welcome!
Try:
codanote.vercel.app
A new field experiment on 515 startups, half shown case studies of how startups are successfully using AI.
Those firms used AI 44% more, had 1.9x higher revenue, needed 39% less capital:
1) AI accelerates businesses
2) The challenge is understanding how to use it papers.ssrn.com/sol3/Deliver...
We must go further: redesign institutions, test causal interventions, and have national funders establish dedicated meta-research panels: an investment this field has long deserved.
📄 Miske et al. doi.org/10.1038/s415...
📄 doi.org/10.1038/d415...
@ucoimbra.bsky.social @excelscior-era.bsky.social
Commentaries on SCORE papers from @asanchez-tojar.bsky.social, Jelte Wicherts, and @robbwiller.bsky.social
www.nature.com/articles/d41...
The more shocking thing here is that academics have been paying for copyediting services given the incomprehensibility of papers. I'm being a bit glib and this is improving over time but it's kind of wild to think people are paying for this when the typical abstract is still so dense and hard to rea
I’m thrilled to see this glowing review of my new book (The Science of Second Chances) from Keith Humphreys in the Washington Monthly.
I’ve long-admired Keith’s work for “factivism” over activism in the public safety space. There is, of course, lots more to do. Onward!
Statistical Rethinking 2026 is done: 20 new lectures emphasizing logical and critical statistical workflow, from basics of probability theory to causal inference to reliable computation to sensitivity. It's all free, made just for you. Lecture list and links: github.com/rmcelreath/s...
AI really can help education: controlled experiment found a GPT-4o powered tutor that personalized problems for students raised final test scores by .15 SD "equivalent to as much as six to nine months of additional schooling by some estimates—without increasing instruction time or teacher workload"
I think this is an important pattern that will become more important. Our society is laden with well-intentioned rules and policies that do limited harm only because the humans don’t implement them faithfully.
A piece by Sheila Heti about consumers' psychology and the different ways we shop
www.affidavit.art/articles/sho...
Old man post incoming: if you are someone who enjoys talking about ideas or lively debate, grab your university years by the horns! One of the greatest gifts of your uni years is the selection effect of being near other smart people who share your interests! #reminiscing
It is one of the weirdest divides, I speak to two companies in the exact same industry and one has been using AI for the past 18 months and the other has a committee that has to approve every use case individually and talk about how AI companies will train on their data.
This is consistent with earlier psychometric work that suggests 5-7 is the best response scale options, but good to see that the finding holds up in contemporary research. Also, good to see that labeling scales whether anchored or not has little impact on findings. academic.oup.com/ijpor/articl...
Not making any sort of claims or trying to imply anything about this one certain study. But for these type of ai studies, I think it's even more important for open data practices to be in place since the exact wording of prompts, system setup etc matter. Replication should be faster too and expected
Every few months, I write an updated, idiosyncratic guide on which AIs to use right now.
My new version has the most changes ever, since AI is no longer just about chatbots. To use AI you need to understand how to think about models, apps, and harnesses. open.substack.com/pub/oneusefu...
We spent months training grad student RAs and GPT-5 mini still beat them by a lot
We have a new pre-print! 📝🖨️
We find that conversing with a disagreeing LLM helped improve people's inaccurate predictions!
osf.io/preprints/ps...
Let me tell you all about it:
Yeah no shade your way. I retweet in the same way.
Given your other posts lately on blue Monday and stuff, it was interesting to see this without the caveat of "hey this may or may not be a real thing".
But yeah realistically it's just me reading another blog post at 4am while watching the baby and getting cranky.
As I said originally, it doesn't seem crazy that there could be some small effect. But this post is not evidence for one in my view.
More thoughts: 6) no statistical tests here. 7) no time series modelling 8) effect for any actual business would be tiny!! 9) February Mondays include presidents Day in US, how is that accounted for? 10) no comparison to other holiday Mondays 11) no comparison to other Mondays for last minute PTO
I feel so many ways about this. 1) seems highly believable. 2) seems like one of those cutesy findings that with more poking into data could disappear. 3) this company only cover 34,000 very non representative companies. 4) effect size is small. 5) this company wants a splashy marketing story
Introducing “Pretend Battleship”: you’re told where all the ships are but then have to play like you never got that information. Could you do it? And what would your performance reveal about your understanding of your own mind? A joy to be part of this creative project led by @matanmazor.bsky.social
Large Language Models have shown both remarkable reasoning ability, and significant reasoning failures.
Research by @psong1.bsky.social et al has categorized this phenomenon in a clear taxonomy to explore how and why the performance is so variable:
buff.ly/hI7DYYN
Thrilled to share our latest paper, out now in Science Advances! We explored the development of cooperative behaviors — fairness, trustworthiness, forgiveness, & honesty — across five societies, culturally contextualizing them & seeing how they correlate. (1/5) www.science.org/doi/full/10....
