Jean-François Bonnefon
@jfbonnefon.bsky.social
Behavioral Science of AI @ Toulouse School of Economics @tse-fr.eu Director of @iast.fr Chair of Moral AI @ Artificial and Natural Intelligence Toulouse Institute https://jfbonnefon.github.io/ I have a successful life with bipolar disorder
Reposted by Jean-François Bonnefon
I mentioned this approach to a colleague who does research on the use of AI in education. She replied, “Guess what some students do in this situation. They prompt ChatGPT to write a report on the subject and an accompanying report explaining how and why the first report is wrong.”
November 24, 2025 at 12:42 PM
I mentioned this approach to a colleague who does research on the use of AI in education. She replied, “Guess what some students do in this situation. They prompt ChatGPT to write a report on the subject and an accompanying report explaining how and why the first report is wrong.”
Me when the Q&A at the seminar goes into overtime
I might have posted this before but I am still looking for a situation where this panel would be relevant as a reaction pic. I haven’t found it yet, so the search continues.
November 23, 2025 at 9:35 PM
Me when the Q&A at the seminar goes into overtime
Did you know that using the words "em dash" in a post is 100% a sign that it was written by AI
November 23, 2025 at 1:16 PM
Did you know that using the words "em dash" in a post is 100% a sign that it was written by AI
Reposted by Jean-François Bonnefon
@sjdm-tweets.bsky.social
Why does "1-in-X" (e.g., 1 in 50 vs. 20 in 1000) feel riskier? The answer: It's the availability heuristic! We found that the 1-in-X ratio is easier to mentally visualize and recall examples for.
Risk Communication Poster at #SJDM2025 in Denver, Colorado!
Why does "1-in-X" (e.g., 1 in 50 vs. 20 in 1000) feel riskier? The answer: It's the availability heuristic! We found that the 1-in-X ratio is easier to mentally visualize and recall examples for.
Risk Communication Poster at #SJDM2025 in Denver, Colorado!
November 23, 2025 at 1:25 AM
@sjdm-tweets.bsky.social
Why does "1-in-X" (e.g., 1 in 50 vs. 20 in 1000) feel riskier? The answer: It's the availability heuristic! We found that the 1-in-X ratio is easier to mentally visualize and recall examples for.
Risk Communication Poster at #SJDM2025 in Denver, Colorado!
Why does "1-in-X" (e.g., 1 in 50 vs. 20 in 1000) feel riskier? The answer: It's the availability heuristic! We found that the 1-in-X ratio is easier to mentally visualize and recall examples for.
Risk Communication Poster at #SJDM2025 in Denver, Colorado!
Reposted by Jean-François Bonnefon
Folks, Zohran Mamdani—smart guy, reminds me of a young me—you know what he said, I told him, “Zohran, I’m gonna do it, I’m gonna say the shahada,” and he said to me “Mr. President, you should say a third line confirming your belief in the wilayat of Ali,” should we do it folks? Should we say it?
November 22, 2025 at 1:31 AM
Folks, Zohran Mamdani—smart guy, reminds me of a young me—you know what he said, I told him, “Zohran, I’m gonna do it, I’m gonna say the shahada,” and he said to me “Mr. President, you should say a third line confirming your belief in the wilayat of Ali,” should we do it folks? Should we say it?
Reposted by Jean-François Bonnefon
Sales of AI-enabled teddy bear suspended after it gave advice on BDSM sex and where to find knives
Sales of AI-enabled teddy bear suspended after it gave advice on BDSM sex and where to find knives | CNN Business
Sales of an artificial intelligence-enabled plush toy have been suspended after it was found that it engaged in conversation around sexually explicit topics and offered potentially dangerous advice.
www.cnn.com
November 20, 2025 at 3:35 AM
Sales of AI-enabled teddy bear suspended after it gave advice on BDSM sex and where to find knives
Reposted by Jean-François Bonnefon
europe has entered the chat 👀
November 21, 2025 at 4:45 AM
europe has entered the chat 👀
Reposted by Jean-François Bonnefon
Having written and thought a bit about tax evasion strategies, I think study 4 in this paper (tax evasion with LLMs) is extremely important. Tax evasion strategies proliferate in places where LLMs have been trained like reddit and guardrails can be tricky where legality of behavior can be fuzzy.
I'd really want to know about the guardrails being implemented here, given our recent results in @nature.com showing that people are likely to ask LLMs to help them cheat with taxes, and that LLMs are likely to accept
www.nature.com/articles/s41...
www.nature.com/articles/s41...
TurboTax is about to be integrated into ChatGPT, where customers will be guided through tasks tied to their tax filings or financial profile by the AI chatbot.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
November 19, 2025 at 1:51 PM
Having written and thought a bit about tax evasion strategies, I think study 4 in this paper (tax evasion with LLMs) is extremely important. Tax evasion strategies proliferate in places where LLMs have been trained like reddit and guardrails can be tricky where legality of behavior can be fuzzy.
I'd really want to know about the guardrails being implemented here, given our recent results in @nature.com showing that people are likely to ask LLMs to help them cheat with taxes, and that LLMs are likely to accept
www.nature.com/articles/s41...
www.nature.com/articles/s41...
TurboTax is about to be integrated into ChatGPT, where customers will be guided through tasks tied to their tax filings or financial profile by the AI chatbot.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
November 19, 2025 at 11:41 AM
I'd really want to know about the guardrails being implemented here, given our recent results in @nature.com showing that people are likely to ask LLMs to help them cheat with taxes, and that LLMs are likely to accept
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Reposted by Jean-François Bonnefon
Apply now to our next Summer School! 🤝
We are glad to host the Toulouse Summer School in Quantitative Social Sciences from 26 May to 19 June 2026 — a great opportunity for PhD students in economics, political science, and other social sciences.
Application: www.tse-fr.eu/toulouse-sum...
We are glad to host the Toulouse Summer School in Quantitative Social Sciences from 26 May to 19 June 2026 — a great opportunity for PhD students in economics, political science, and other social sciences.
Application: www.tse-fr.eu/toulouse-sum...
November 19, 2025 at 9:17 AM
Apply now to our next Summer School! 🤝
We are glad to host the Toulouse Summer School in Quantitative Social Sciences from 26 May to 19 June 2026 — a great opportunity for PhD students in economics, political science, and other social sciences.
Application: www.tse-fr.eu/toulouse-sum...
We are glad to host the Toulouse Summer School in Quantitative Social Sciences from 26 May to 19 June 2026 — a great opportunity for PhD students in economics, political science, and other social sciences.
Application: www.tse-fr.eu/toulouse-sum...
What is the common interpretation when PIs (in psychology) list themselves as 1st authors of papers co-published with their students/postdocs, and the order is not alphabetical? Clearly they are not applying the first/last author convention, so do they mean that they did most of the work?
November 17, 2025 at 2:05 PM
What is the common interpretation when PIs (in psychology) list themselves as 1st authors of papers co-published with their students/postdocs, and the order is not alphabetical? Clearly they are not applying the first/last author convention, so do they mean that they did most of the work?
I think we are shifting to a better timeline because what is the other explanation for Charlie xcx doing a Wuthering Heights soundtrack inspired by the Velvet Underground?
November 16, 2025 at 9:18 AM
I think we are shifting to a better timeline because what is the other explanation for Charlie xcx doing a Wuthering Heights soundtrack inspired by the Velvet Underground?
Reposted by Jean-François Bonnefon
I know my goth card will get revoked but... I have never seen "Interview with a Vampire" 😞
November 13, 2025 at 11:42 PM
I know my goth card will get revoked but... I have never seen "Interview with a Vampire" 😞
❝Given the quality of the manuscripts you submitted to our journal, we believe you are uniquely qualified to identify work that does not pass our bar for rigor and novelty❞
Why should you accept invitations to review from journals that have rejected every single one of your submissions?
November 13, 2025 at 2:05 PM
❝Given the quality of the manuscripts you submitted to our journal, we believe you are uniquely qualified to identify work that does not pass our bar for rigor and novelty❞
I would be remiss if I didn't take this opportunity to link to our newest paper in @natclimate.nature.com :
❝Using generative AI to increase sceptics' engagement with climate science❞
www.nature.com/articles/s41...
❝Using generative AI to increase sceptics' engagement with climate science❞
www.nature.com/articles/s41...
Countries commit to tackling climate disinformation at COP30
It is the first time states have formally committed to information integrity and fighting back against climate disinformation.
www.euronews.com
November 13, 2025 at 1:02 PM
I would be remiss if I didn't take this opportunity to link to our newest paper in @natclimate.nature.com :
❝Using generative AI to increase sceptics' engagement with climate science❞
www.nature.com/articles/s41...
❝Using generative AI to increase sceptics' engagement with climate science❞
www.nature.com/articles/s41...
Sitting just a few meters from Macron at a town hall ‘debate’. Not sure it counts as a debate when he speaks nonstop, but his command of language is so impressive. It’s quite a feat to remain precise and engaging all through when you give 30-minute answers.
November 12, 2025 at 11:34 AM
Sitting just a few meters from Macron at a town hall ‘debate’. Not sure it counts as a debate when he speaks nonstop, but his command of language is so impressive. It’s quite a feat to remain precise and engaging all through when you give 30-minute answers.
Reposted by Jean-François Bonnefon
crazy how much this bugs me now that someone pointed it out. if i worked at the white house, i would find this ugly, but not as annoying as i do now. ignorance is really bliss bc they more you learn, the more you just walk around thinking "collar gap" and "kerning is off."
November 12, 2025 at 6:44 AM
crazy how much this bugs me now that someone pointed it out. if i worked at the white house, i would find this ugly, but not as annoying as i do now. ignorance is really bliss bc they more you learn, the more you just walk around thinking "collar gap" and "kerning is off."
Reposted by Jean-François Bonnefon
Okay, so fantasy football is a thing.
But doesn't that imply the existence of mystery football, horror football and the like? Maybe Regency football, screwball football, Ruritanian football and more?
But doesn't that imply the existence of mystery football, horror football and the like? Maybe Regency football, screwball football, Ruritanian football and more?
November 11, 2025 at 4:09 AM
Okay, so fantasy football is a thing.
But doesn't that imply the existence of mystery football, horror football and the like? Maybe Regency football, screwball football, Ruritanian football and more?
But doesn't that imply the existence of mystery football, horror football and the like? Maybe Regency football, screwball football, Ruritanian football and more?
Reposted by Jean-François Bonnefon
woah plot twist
November 10, 2025 at 7:42 AM
woah plot twist
Reposted by Jean-François Bonnefon
This is a cool paper showing that first-gen college students don't realize a lot of unwritten rules that lead to success (the value of internships, student clubs, letters from professors).
But giving them access to an LLM for guidance significantly closes the gap. mgcuna.github.io/website/JMP_...
But giving them access to an LLM for guidance significantly closes the gap. mgcuna.github.io/website/JMP_...
November 9, 2025 at 2:55 PM
This is a cool paper showing that first-gen college students don't realize a lot of unwritten rules that lead to success (the value of internships, student clubs, letters from professors).
But giving them access to an LLM for guidance significantly closes the gap. mgcuna.github.io/website/JMP_...
But giving them access to an LLM for guidance significantly closes the gap. mgcuna.github.io/website/JMP_...
French Theory, the collectible card game edition. I got a Kristeva and a Lacan and they have ok powers, but good luck getting a Foucault or a Bourdieu
November 9, 2025 at 3:11 PM
French Theory, the collectible card game edition. I got a Kristeva and a Lacan and they have ok powers, but good luck getting a Foucault or a Bourdieu
It can be hard for multidisciplinary teams to agree on the title of their manuscripts
November 8, 2025 at 12:33 PM
It can be hard for multidisciplinary teams to agree on the title of their manuscripts
Please select six keywords for your manuscript
November 8, 2025 at 11:53 AM
Please select six keywords for your manuscript
Reposted by Jean-François Bonnefon
🚨Paper Alert 🚨 Did 1st contact with ChatGPT alter sentiment about AI?
First interactions with generative chatbots shape local but not global sentiments about AI
Work with Eva Schmidt, Mengchen Dong, @jfbonnefon.bsky.social, Clara Bersch, @nckobis.bsky.social
www.sciencedirect.com/science/arti...
First interactions with generative chatbots shape local but not global sentiments about AI
Work with Eva Schmidt, Mengchen Dong, @jfbonnefon.bsky.social, Clara Bersch, @nckobis.bsky.social
www.sciencedirect.com/science/arti...
First interactions with generative chatbots shape local but not global sentiments about AI
As artificial intelligence (AI) chatbots become increasingly integrated into everyday life, it is important to understand how direct interaction with …
www.sciencedirect.com
November 6, 2025 at 5:03 PM
🚨Paper Alert 🚨 Did 1st contact with ChatGPT alter sentiment about AI?
First interactions with generative chatbots shape local but not global sentiments about AI
Work with Eva Schmidt, Mengchen Dong, @jfbonnefon.bsky.social, Clara Bersch, @nckobis.bsky.social
www.sciencedirect.com/science/arti...
First interactions with generative chatbots shape local but not global sentiments about AI
Work with Eva Schmidt, Mengchen Dong, @jfbonnefon.bsky.social, Clara Bersch, @nckobis.bsky.social
www.sciencedirect.com/science/arti...
oh hi darkness my old friend
November 4, 2025 at 10:22 AM
oh hi darkness my old friend