M.J. Crockett
@mjcrockett.bsky.social
Professor of Psychology & Human Values at Princeton | Cognitive scientist curious about technology, narratives, & epistemic (in)justice | They/She 🏳️🌈
www.crockettlab.org
www.crockettlab.org
Pinned
M.J. Crockett
@mjcrockett.bsky.social
· Oct 21
AI Surrogates and illusions of generalizability in cognitive science
Recent advances in artificial intelligence (AI) have generated enthusiasm for using AI simulations of human research participants to generate new know…
www.sciencedirect.com
Can AI simulations of human research participants advance cognitive science? In @cp-trendscognsci.bsky.social, @lmesseri.bsky.social & I analyze this vision. We show how “AI Surrogates” entrench practices that limit the generalizability of cognitive science while aspiring to do the opposite. 1/
Reposted by M.J. Crockett
I'm shocked no-one has started cosplaying Dreyfus on AI yet! philpapers.org/rec/DREWHA But I also found this ace piece on automation from the early 80s last week which is like every ai in work application www.sciencedirect.com/science/arti...
Ironies of automation
This paper discusses the ways in which automation of industrial processes may expand rather than eliminate problems with the human operator. Some comm…
www.sciencedirect.com
December 3, 2025 at 1:10 PM
I'm shocked no-one has started cosplaying Dreyfus on AI yet! philpapers.org/rec/DREWHA But I also found this ace piece on automation from the early 80s last week which is like every ai in work application www.sciencedirect.com/science/arti...
Reposted by M.J. Crockett
December 3, 2025 at 1:29 PM
Reposted by M.J. Crockett
So many things.
Have you read Philip Agre's "Toward a Critical Technical Practice" ? pages.gseis.ucla.edu/faculty/agre...
Have you read Philip Agre's "Toward a Critical Technical Practice" ? pages.gseis.ucla.edu/faculty/agre...
pages.gseis.ucla.edu
December 3, 2025 at 3:12 AM
So many things.
Have you read Philip Agre's "Toward a Critical Technical Practice" ? pages.gseis.ucla.edu/faculty/agre...
Have you read Philip Agre's "Toward a Critical Technical Practice" ? pages.gseis.ucla.edu/faculty/agre...
What are your favorite “vintage” pieces on AI that are just as relevant today?
It’s wild reading critical scholarship on AI from 20-30 years ago and seeing how little has changed.
December 3, 2025 at 1:03 PM
What are your favorite “vintage” pieces on AI that are just as relevant today?
Reposted by M.J. Crockett
Cannot second this one enough! If you haven't read it, I'd also recommend John Pierce's Whither Speech Recognition, which I think has some interesting resonances today, and some early inklings of how a lot of later technology ended up working out, couched in so.e *pure* venom
Whither Speech Recognition?
Speech recognition has glamour. Funds have been available. Results have been less glamorous. “When we listen to a person speaking much of what we think we hear
pubs.aip.org
December 3, 2025 at 4:48 AM
Cannot second this one enough! If you haven't read it, I'd also recommend John Pierce's Whither Speech Recognition, which I think has some interesting resonances today, and some early inklings of how a lot of later technology ended up working out, couched in so.e *pure* venom
It’s wild reading critical scholarship on AI from 20-30 years ago and seeing how little has changed.
December 3, 2025 at 3:05 AM
It’s wild reading critical scholarship on AI from 20-30 years ago and seeing how little has changed.
Reposted by M.J. Crockett
All our work on replicability in the last decade has taught me that it's not a good use of resources to try and predict replicability in the first place since it's neither a good goal to achieve nor diagnostic of specific problems. It's a search for shortcuts, easy answers to complex questions.
There are many good arguments for why AI/ML will struggle to predict replicability.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
December 3, 2025 at 2:29 AM
All our work on replicability in the last decade has taught me that it's not a good use of resources to try and predict replicability in the first place since it's neither a good goal to achieve nor diagnostic of specific problems. It's a search for shortcuts, easy answers to complex questions.
Reposted by M.J. Crockett
I had the privilege of delivering my first talk at @princetoncitp.bsky.social today.
December 2, 2025 at 8:02 PM
I had the privilege of delivering my first talk at @princetoncitp.bsky.social today.
There are many good arguments for why AI/ML will struggle to predict replicability.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
December 2, 2025 at 6:58 PM
There are many good arguments for why AI/ML will struggle to predict replicability.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
www.pnas.org/doi/full/10....
I'm curious why, in light of these arguments and now data, COS is continuing to pursue this project.
Reposted by M.J. Crockett
This cross-pollination IS how new ideas happen. EVERY field needs humanistic thinking, that's why they call PhDs "Doctors of Philosophy"
Alt text for those who need it. This is amazing.
December 2, 2025 at 11:38 AM
This cross-pollination IS how new ideas happen. EVERY field needs humanistic thinking, that's why they call PhDs "Doctors of Philosophy"
Reposted by M.J. Crockett
My wife has been HIV+ for 34 years. She’s alive today b/c work of activists who destigmatized the virus, pushed medical science to adapt & find ways to treat it, & advocated for gov health programs that helped fund her care. The current U.S. regime may want to go back in time, but we won’t let them.
December 2, 2025 at 1:27 AM
My wife has been HIV+ for 34 years. She’s alive today b/c work of activists who destigmatized the virus, pushed medical science to adapt & find ways to treat it, & advocated for gov health programs that helped fund her care. The current U.S. regime may want to go back in time, but we won’t let them.
Reposted by M.J. Crockett
It's that time of year again
December 1, 2025 at 1:23 PM
It's that time of year again
Reposted by M.J. Crockett
A quick thread on the academic freedom angle to the OU situation:
First, grading is part of academic freedom b/c it involves instructors applying their disciplinary expertise - both when they set assignments, and when they grade student work
First, grading is part of academic freedom b/c it involves instructors applying their disciplinary expertise - both when they set assignments, and when they grade student work
December 1, 2025 at 6:19 PM
A quick thread on the academic freedom angle to the OU situation:
First, grading is part of academic freedom b/c it involves instructors applying their disciplinary expertise - both when they set assignments, and when they grade student work
First, grading is part of academic freedom b/c it involves instructors applying their disciplinary expertise - both when they set assignments, and when they grade student work
Reposted by M.J. Crockett
It’s useful to think about things like this in numbers of tenure track professors.
$33M is 358 assistant professors. Or about fourteen percent of the total number of full-time faculty at Michigan State.
$33M is 358 assistant professors. Or about fourteen percent of the total number of full-time faculty at Michigan State.
SOURCES: Michigan State is firing head coach Jonathan Smith, @theathletic.com has learned.
He went 9-15 in two seasons, including 1-8 in the Big Ten this year. (ESPN first reported).
MSU players meeting called for 2:45. Smith has a $33 million buyout (with offset).
He went 9-15 in two seasons, including 1-8 in the Big Ten this year. (ESPN first reported).
MSU players meeting called for 2:45. Smith has a $33 million buyout (with offset).
December 1, 2025 at 12:05 AM
It’s useful to think about things like this in numbers of tenure track professors.
$33M is 358 assistant professors. Or about fourteen percent of the total number of full-time faculty at Michigan State.
$33M is 358 assistant professors. Or about fourteen percent of the total number of full-time faculty at Michigan State.
Reposted by M.J. Crockett
Many, many unpaid labour-hours will be required for the ICLR authors and organizers to sort this mess out, and the conference will be worse for it. A perfect case study on why AI productivity gains are largely a mirage.
November 30, 2025 at 7:01 PM
Many, many unpaid labour-hours will be required for the ICLR authors and organizers to sort this mess out, and the conference will be worse for it. A perfect case study on why AI productivity gains are largely a mirage.
"The ICLR 2026 team permitted authors and reviewers to use AI tools to polish text, generate experiment codes or analyse results, but mandated disclosure of such uses."
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
Major AI conference flooded with peer reviews written fully by AI
Controversy has erupted after 21% of manuscript reviews for an international AI conference were found to be generated by artificial intelligence.
www.nature.com
November 30, 2025 at 5:20 PM
"The ICLR 2026 team permitted authors and reviewers to use AI tools to polish text, generate experiment codes or analyse results, but mandated disclosure of such uses."
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
Great example of how policies that allow *any* use of AI in writing and reviewing readily give way to wholesale fabrication.
Thoughtful and thought-provoking!
What makes something data? Some thoughts on that question, and how answers to it help us understand AI hype:
medium.com/@emilymenonb...
medium.com/@emilymenonb...
What makes something data?
This is a question I posted on BlueSky on Friday 11/21/25, inspired by a talk I recently attended about evaluation of “AI” systems. I think…
medium.com
November 29, 2025 at 10:33 PM
Thoughtful and thought-provoking!
Reposted by M.J. Crockett
If you're interested, this tribute to Alice Wong was written by one of the people who knew her best
www.thenation.com/article/soci...
www.thenation.com/article/soci...
A Tribute to an Oracle, Alice Wong
Alice had the ability to look to the future and a world where laws and attitudes did not keep disabled people poor, pitied, and isolated.
www.thenation.com
November 26, 2025 at 11:18 PM
If you're interested, this tribute to Alice Wong was written by one of the people who knew her best
www.thenation.com/article/soci...
www.thenation.com/article/soci...
Reposted by M.J. Crockett
On AI’s ‘mediocrity trap’ — experiments indicates that while AI helps the less skilled make something passable, the highly skilled don’t use it to produce something better than they could have; they produce something ok, but lose motivation to make it great. www.jin-li.org/uploads/1/1/...
November 26, 2025 at 9:18 PM
On AI’s ‘mediocrity trap’ — experiments indicates that while AI helps the less skilled make something passable, the highly skilled don’t use it to produce something better than they could have; they produce something ok, but lose motivation to make it great. www.jin-li.org/uploads/1/1/...
Reposted by M.J. Crockett
"Actually, you harmed us by using our chatbot in a way that made us look bad" is straight out of the abuser's handbook
ChatGPT firm blames boy’s suicide on ‘misuse’ of its technology
OpenAI responds to lawsuit claiming its chatbot encouraged California teenager to kill himself
www.theguardian.com
November 26, 2025 at 8:54 PM
"Actually, you harmed us by using our chatbot in a way that made us look bad" is straight out of the abuser's handbook
Reposted by M.J. Crockett
journalists really should stop baring their entire asses like this
"I have no idea how this technology actually works but my editor is mostly interested in engagement clicks so we proceeded anyway"
"I have no idea how this technology actually works but my editor is mostly interested in engagement clicks so we proceeded anyway"
November 26, 2025 at 6:41 PM
journalists really should stop baring their entire asses like this
"I have no idea how this technology actually works but my editor is mostly interested in engagement clicks so we proceeded anyway"
"I have no idea how this technology actually works but my editor is mostly interested in engagement clicks so we proceeded anyway"
Reposted by M.J. Crockett
If you are interested in the methodology behind the Data Workers' Inquiry Project (@dataworkersinquiry.bsky.social), this is the paper for you.
Research doesn't have to be exploitative.👇🏽
Research doesn't have to be exploitative.👇🏽
📣New paper!
Thrilled to share the methodological recipe for @dataworkersinquiry.bsky.social in this article co-authored with @adiod.bsky.social @krystalkauffman.bsky.social Camilla Salim Wagner, Laurenz Sachenbacher @alexhanna.bsky.social & @timnitgebru.bsky.social
👉 ojs.aaai.org/index.php/AI...
Thrilled to share the methodological recipe for @dataworkersinquiry.bsky.social in this article co-authored with @adiod.bsky.social @krystalkauffman.bsky.social Camilla Salim Wagner, Laurenz Sachenbacher @alexhanna.bsky.social & @timnitgebru.bsky.social
👉 ojs.aaai.org/index.php/AI...
November 26, 2025 at 12:45 AM
If you are interested in the methodology behind the Data Workers' Inquiry Project (@dataworkersinquiry.bsky.social), this is the paper for you.
Research doesn't have to be exploitative.👇🏽
Research doesn't have to be exploitative.👇🏽
Amazing news!!!
Honored to serve on Mayor-elect @zohrankmamdani.bsky.social Transition Committee on Technology. For too long, too many New Yorkers have been left behind. NYC deserves governance that centers affordability, equity, and the public good-looking forward to the work ahead. www.cbsnews.com/newyork/news...
NYC Mayor-elect Zohran Mamdani announces transition committees
New York City Mayor-elect Zohran Mamdani announced Monday the creation of 17 transition advisory committees made up of more than 400 people.
www.cbsnews.com
November 24, 2025 at 11:30 PM
Amazing news!!!
Reposted by M.J. Crockett
Meta, my expectations for you were low, but JFC
November 23, 2025 at 8:49 PM
Meta, my expectations for you were low, but JFC
Reposted by M.J. Crockett
If you're trying to convince your friends to spend time on Bluesky rather than X, maybe show them this. X is not "the public town square" or about "free speech." It's "just what Elon Musk wants you to see." If you see stuff he doesn't like he'll step in to make sure you don't see it for long.
If you scroll on X, you are only being shown what Elon wants you to see. If you create content on X, no one will see it unless Elon wants them to.
November 21, 2025 at 1:31 AM
If you're trying to convince your friends to spend time on Bluesky rather than X, maybe show them this. X is not "the public town square" or about "free speech." It's "just what Elon Musk wants you to see." If you see stuff he doesn't like he'll step in to make sure you don't see it for long.