Tom Costello
@tomcostello.bsky.social
research psychologist. beliefs, AI, computational social science. assistant prof at Carnegie Mellon
Reposted by Tom Costello
Episode 2 of On Wisdom's five-part special series is live!
This week features a discussion on Metacognition and Judgment with hosts Igor Grossmann and Charles Cassidy. They are joined by guests Gord Pennycook, Steve Fleming, and Thomas Costello.
🎧Listen here: www.onwisdompodcast.com/70
This week features a discussion on Metacognition and Judgment with hosts Igor Grossmann and Charles Cassidy. They are joined by guests Gord Pennycook, Steve Fleming, and Thomas Costello.
🎧Listen here: www.onwisdompodcast.com/70
October 8, 2026 at 5:39 PM
Episode 2 of On Wisdom's five-part special series is live!
This week features a discussion on Metacognition and Judgment with hosts Igor Grossmann and Charles Cassidy. They are joined by guests Gord Pennycook, Steve Fleming, and Thomas Costello.
🎧Listen here: www.onwisdompodcast.com/70
This week features a discussion on Metacognition and Judgment with hosts Igor Grossmann and Charles Cassidy. They are joined by guests Gord Pennycook, Steve Fleming, and Thomas Costello.
🎧Listen here: www.onwisdompodcast.com/70
More from our group on this! (And using the same cool format as Matt):
the effect comes mostly from facts and evidence¹ and holds for conspiracies only days old².
LLMs can also talk people into conspiracies³.
¹ psyarxiv.com/h7n8u
² arxiv.org/abs/2608.06151
³ arxiv.org/abs/2601.05050
the effect comes mostly from facts and evidence¹ and holds for conspiracies only days old².
LLMs can also talk people into conspiracies³.
¹ psyarxiv.com/h7n8u
² arxiv.org/abs/2608.06151
³ arxiv.org/abs/2601.05050
October 4, 2026 at 6:28 PM
More from our group on this! (And using the same cool format as Matt):
the effect comes mostly from facts and evidence¹ and holds for conspiracies only days old².
LLMs can also talk people into conspiracies³.
¹ psyarxiv.com/h7n8u
² arxiv.org/abs/2608.06151
³ arxiv.org/abs/2601.05050
the effect comes mostly from facts and evidence¹ and holds for conspiracies only days old².
LLMs can also talk people into conspiracies³.
¹ psyarxiv.com/h7n8u
² arxiv.org/abs/2608.06151
³ arxiv.org/abs/2601.05050
Tomorrow morning in DC I'm moderating a panel on feeds and politics with @cerenbudak.bsky.social, @filipecampante.bsky.social and @ylelkes.bsky.social. Misinformation, politicians writing for the algorithm, why depolarization is so hard, and what changes when people ask gen AI instead.
September 28, 2026 at 11:04 PM
Tomorrow morning in DC I'm moderating a panel on feeds and politics with @cerenbudak.bsky.social, @filipecampante.bsky.social and @ylelkes.bsky.social. Misinformation, politicians writing for the algorithm, why depolarization is so hard, and what changes when people ask gen AI instead.
Reposted by Tom Costello
Happy to share a new preprint with
@tomcostello.bsky.social, @dgrand.bsky.social and @gordpennycook.bsky.social 🧵
"What beliefs are most debunkable?" osf.io/preprints/ps...
We investigate the characteristics of the beliefs that are (or are not) influenced by counterevidence in 3,341 AI dialogues.
@tomcostello.bsky.social, @dgrand.bsky.social and @gordpennycook.bsky.social 🧵
"What beliefs are most debunkable?" osf.io/preprints/ps...
We investigate the characteristics of the beliefs that are (or are not) influenced by counterevidence in 3,341 AI dialogues.
September 21, 2026 at 7:31 PM
Happy to share a new preprint with
@tomcostello.bsky.social, @dgrand.bsky.social and @gordpennycook.bsky.social 🧵
"What beliefs are most debunkable?" osf.io/preprints/ps...
We investigate the characteristics of the beliefs that are (or are not) influenced by counterevidence in 3,341 AI dialogues.
@tomcostello.bsky.social, @dgrand.bsky.social and @gordpennycook.bsky.social 🧵
"What beliefs are most debunkable?" osf.io/preprints/ps...
We investigate the characteristics of the beliefs that are (or are not) influenced by counterevidence in 3,341 AI dialogues.
Reposted by Tom Costello
AI Safety
Ask any question. Get a strong explanation.
debunkbot.com
September 10, 2026 at 8:50 PM
Reposted by Tom Costello
Reposted by Tom Costello
Conditionally accepted at the APSR (w/ @scottclifford.bsky.social & @patrickpliu.bsky.social):
Why does political information so often change beliefs but NOT attitudes? We highlight the role of belief relevance, or the extent to which beliefs bear on attitudes.
Why does political information so often change beliefs but NOT attitudes? We highlight the role of belief relevance, or the extent to which beliefs bear on attitudes.
March 9, 2026 at 2:15 PM
Conditionally accepted at the APSR (w/ @scottclifford.bsky.social & @patrickpliu.bsky.social):
Why does political information so often change beliefs but NOT attitudes? We highlight the role of belief relevance, or the extent to which beliefs bear on attitudes.
Why does political information so often change beliefs but NOT attitudes? We highlight the role of belief relevance, or the extent to which beliefs bear on attitudes.
Reposted by Tom Costello
Today's SJDM Featured Paper is:
Costello, T. H., Pelrine, K., Kowal, M., Arechar, A. A., Godbout, J.-F., Gleave, A., Rand, D., & Pennycook, G. (2026). Large language models can effectively convince people to believe conspiracies. arXiv. doi.org/10.48550/arX...
Costello, T. H., Pelrine, K., Kowal, M., Arechar, A. A., Godbout, J.-F., Gleave, A., Rand, D., & Pennycook, G. (2026). Large language models can effectively convince people to believe conspiracies. arXiv. doi.org/10.48550/arX...
Large language models can effectively convince people to believe conspiracies
Large language models (LLMs) have been shown to be persuasive across a variety of contexts. But it remains unclear whether this persuasive power advantages truth over falsehood, or if LLMs can promote...
doi.org
March 5, 2026 at 1:58 PM
Today's SJDM Featured Paper is:
Costello, T. H., Pelrine, K., Kowal, M., Arechar, A. A., Godbout, J.-F., Gleave, A., Rand, D., & Pennycook, G. (2026). Large language models can effectively convince people to believe conspiracies. arXiv. doi.org/10.48550/arX...
Costello, T. H., Pelrine, K., Kowal, M., Arechar, A. A., Godbout, J.-F., Gleave, A., Rand, D., & Pennycook, G. (2026). Large language models can effectively convince people to believe conspiracies. arXiv. doi.org/10.48550/arX...
I saw we were fighting about what AI will do to (social) science.
I have some thoughts (originally written for a book on scientific inquiry) and decided to turn them into my first substack article. Co-written with Nat Rabb!
tomstello.substack.com/p/standing-o...
I have some thoughts (originally written for a book on scientific inquiry) and decided to turn them into my first substack article. Co-written with Nat Rabb!
tomstello.substack.com/p/standing-o...
Standing on the Shoulders of Termites
Or, how we learned to stop worrying and love AI (scientists)
tomstello.substack.com
March 5, 2026 at 3:29 PM
I saw we were fighting about what AI will do to (social) science.
I have some thoughts (originally written for a book on scientific inquiry) and decided to turn them into my first substack article. Co-written with Nat Rabb!
tomstello.substack.com/p/standing-o...
I have some thoughts (originally written for a book on scientific inquiry) and decided to turn them into my first substack article. Co-written with Nat Rabb!
tomstello.substack.com/p/standing-o...
Reposted by Tom Costello
🧵on my new paper "Synthetic personas distort the structure of human belief systems" w Roberto Cerina I'm v excited about...
🚨 Do synthetic samples look like human samples?
We compare 28 LLMs to the 2024 General Social Survey (GSS) to find out + develop host of diagnostics...
🚨 Do synthetic samples look like human samples?
We compare 28 LLMs to the 2024 General Social Survey (GSS) to find out + develop host of diagnostics...
February 25, 2026 at 7:46 PM
🧵on my new paper "Synthetic personas distort the structure of human belief systems" w Roberto Cerina I'm v excited about...
🚨 Do synthetic samples look like human samples?
We compare 28 LLMs to the 2024 General Social Survey (GSS) to find out + develop host of diagnostics...
🚨 Do synthetic samples look like human samples?
We compare 28 LLMs to the 2024 General Social Survey (GSS) to find out + develop host of diagnostics...
Reposted by Tom Costello
I'm hiring a postdoc at @cmu.edu (w/ far.ai & @dgrand.bsky.social + @gordpennycook.bsky.social)!
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
February 23, 2026 at 6:46 PM
I'm hiring a postdoc at @cmu.edu (w/ far.ai & @dgrand.bsky.social + @gordpennycook.bsky.social)!
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
I'm hiring a postdoc at @cmu.edu (w/ far.ai & @dgrand.bsky.social + @gordpennycook.bsky.social)!
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
February 23, 2026 at 6:46 PM
I'm hiring a postdoc at @cmu.edu (w/ far.ai & @dgrand.bsky.social + @gordpennycook.bsky.social)!
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
How do LLMs shape human beliefs — and what do we do about it? AI safety meets behavioral science.
Open to technical and social science backgrounds.
Reposted by Tom Costello
🚨 New paper out at @ajpseditor.bsky.social 🚨
Do the public hold meaningful attitudes? Using the case of abortion policy preferences, we provide strong evidence that policy prefrences can be coherent, stable over time, and causally explain vote choice.
doi.org/10.1111/ajps...
Do the public hold meaningful attitudes? Using the case of abortion policy preferences, we provide strong evidence that policy prefrences can be coherent, stable over time, and causally explain vote choice.
doi.org/10.1111/ajps...
February 18, 2026 at 11:26 PM
🚨 New paper out at @ajpseditor.bsky.social 🚨
Do the public hold meaningful attitudes? Using the case of abortion policy preferences, we provide strong evidence that policy prefrences can be coherent, stable over time, and causally explain vote choice.
doi.org/10.1111/ajps...
Do the public hold meaningful attitudes? Using the case of abortion policy preferences, we provide strong evidence that policy prefrences can be coherent, stable over time, and causally explain vote choice.
doi.org/10.1111/ajps...
Reposted by Tom Costello
New paper in Current Directions in Psych Science: journals.sagepub.com/doi/10.1177/...
After countless arguments about what tasks ppl should/should not offload to AI, we instead argue that genAI can be used *augment* research protocols in novel ways. I.e. use AI to make better psych experiments!
After countless arguments about what tasks ppl should/should not offload to AI, we instead argue that genAI can be used *augment* research protocols in novel ways. I.e. use AI to make better psych experiments!
February 18, 2026 at 10:22 PM
New paper in Current Directions in Psych Science: journals.sagepub.com/doi/10.1177/...
After countless arguments about what tasks ppl should/should not offload to AI, we instead argue that genAI can be used *augment* research protocols in novel ways. I.e. use AI to make better psych experiments!
After countless arguments about what tasks ppl should/should not offload to AI, we instead argue that genAI can be used *augment* research protocols in novel ways. I.e. use AI to make better psych experiments!
Honored (and genuinely, wildly grateful and -- even more than I am grateful -- surprised) to share that our paper "Durably reducing conspiracy beliefs through dialogues with AI" received the @aaas.org Newcomb-Cleveland Prize (for the "most outstanding" paper in @science.org last year).
February 13, 2026 at 11:21 PM
Honored (and genuinely, wildly grateful and -- even more than I am grateful -- surprised) to share that our paper "Durably reducing conspiracy beliefs through dialogues with AI" received the @aaas.org Newcomb-Cleveland Prize (for the "most outstanding" paper in @science.org last year).
Reposted by Tom Costello
🚨New WP: Can an AI voter guide (grounded in information from a nonpartisan, fact-checked source) help voters’ decision making? 🚨
We built and evaluated an LLM-based chatbot that provided voting info in CA & TX (N=2,474) right before the 2024 election. 🧵👇
We built and evaluated an LLM-based chatbot that provided voting info in CA & TX (N=2,474) right before the 2024 election. 🧵👇
February 9, 2026 at 8:56 PM
🚨New WP: Can an AI voter guide (grounded in information from a nonpartisan, fact-checked source) help voters’ decision making? 🚨
We built and evaluated an LLM-based chatbot that provided voting info in CA & TX (N=2,474) right before the 2024 election. 🧵👇
We built and evaluated an LLM-based chatbot that provided voting info in CA & TX (N=2,474) right before the 2024 election. 🧵👇
Reposted by Tom Costello
If the medium is the message, then the message of algorithmic content platforms is not expression or connection or freedom. It is salience above all else.
This is what is happening to us on these platforms. They cannot behave any other way.
This is what is happening to us on these platforms. They cannot behave any other way.
January 28, 2026 at 5:29 PM
If the medium is the message, then the message of algorithmic content platforms is not expression or connection or freedom. It is salience above all else.
This is what is happening to us on these platforms. They cannot behave any other way.
This is what is happening to us on these platforms. They cannot behave any other way.
Reposted by Tom Costello
The #1 New York Times bestseller The Sirens’ Call is now in paperback! Chris Hayes masterfully explores how our focus has been commodified & manipulated, urging us to reclaim control over our lives & future. The Sirens’ Call is the big-picture vision we urgently need to offer clarity & guidance.
January 27, 2026 at 4:22 PM
The #1 New York Times bestseller The Sirens’ Call is now in paperback! Chris Hayes masterfully explores how our focus has been commodified & manipulated, urging us to reclaim control over our lives & future. The Sirens’ Call is the big-picture vision we urgently need to offer clarity & guidance.
Reposted by Tom Costello
After years in academia, I’m exploring data science and research roles in industry.
I'm a quant. social scientist (PhD Yale ’24, NYU) focused on causal inference, experiments, and large-scale data.
Feel free to get in touch or share; all leads appreciated. [email protected]
I'm a quant. social scientist (PhD Yale ’24, NYU) focused on causal inference, experiments, and large-scale data.
Feel free to get in touch or share; all leads appreciated. [email protected]
January 27, 2026 at 6:45 PM
After years in academia, I’m exploring data science and research roles in industry.
I'm a quant. social scientist (PhD Yale ’24, NYU) focused on causal inference, experiments, and large-scale data.
Feel free to get in touch or share; all leads appreciated. [email protected]
I'm a quant. social scientist (PhD Yale ’24, NYU) focused on causal inference, experiments, and large-scale data.
Feel free to get in touch or share; all leads appreciated. [email protected]
Reposted by Tom Costello
Our latest collaboration with CMU, Cornell, and others: AI is just as effective at spreading conspiracy beliefs as it is at debunking them. We found a fix that works, showing that we need to make deliberate design choices. Full thread below.
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
January 20, 2026 at 3:11 PM
Our latest collaboration with CMU, Cornell, and others: AI is just as effective at spreading conspiracy beliefs as it is at debunking them. We found a fix that works, showing that we need to make deliberate design choices. Full thread below.
Reposted by Tom Costello
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
January 20, 2026 at 3:00 PM
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
January 20, 2026 at 3:00 PM
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
Reposted by Tom Costello
A very interesting data-rich analysis of persuasion on digital media by @benmtappin.bsky.social. Recommend!
open.substack.com/pub/benmtapp...
open.substack.com/pub/benmtapp...
For Digital Mass Persuasion, Exposure Matters More Than Persuasiveness
If you are interested in understanding the mass persuasive impact of digital media content you should generally pay more attention to content exposure than to its persuasiveness.
open.substack.com
December 22, 2025 at 1:08 PM
A very interesting data-rich analysis of persuasion on digital media by @benmtappin.bsky.social. Recommend!
open.substack.com/pub/benmtapp...
open.substack.com/pub/benmtapp...
Reposted by Tom Costello
in 2025 the combination of the latest models and agentic harnesses like Claude Code (currently considered tools for coders) felt like a step change for software engineers and, it seems, for mathematicians as well. in 2026 this will hit every other white collar job
xcancel.com/Aaroth/statu...
xcancel.com/Aaroth/statu...
December 22, 2025 at 2:46 PM
in 2025 the combination of the latest models and agentic harnesses like Claude Code (currently considered tools for coders) felt like a step change for software engineers and, it seems, for mathematicians as well. in 2026 this will hit every other white collar job
xcancel.com/Aaroth/statu...
xcancel.com/Aaroth/statu...
Reposted by Tom Costello
I’m less polite: the capabilities q is usually an active denial. in skilled hands, SOTA LLM tools are more capable today than most people think. assume they can do X and you can get through to the real question of what people will do with that and what that reveals about the pressures they’re under
December 14, 2025 at 4:37 PM
I’m less polite: the capabilities q is usually an active denial. in skilled hands, SOTA LLM tools are more capable today than most people think. assume they can do X and you can get through to the real question of what people will do with that and what that reveals about the pressures they’re under