iiinova — where your digital self keeps evolving

Antoine Marie PhD @a-marie-sci.bsky.social · 18/09/2026

𝗛𝗼𝘄 𝗙𝗲𝗱𝗼𝗿𝗼𝘃𝗮 𝗮𝗻𝗱 𝗥𝘂𝘀𝘀𝗶𝗮𝗻 𝗣𝗿𝗼𝗽𝗮𝗴𝗮𝗻𝗱𝗮 𝗧𝘂𝗿𝗻 𝗼𝘂𝗿 𝗚𝗼𝗮𝗹 𝗼𝗳 𝗕𝗲𝗶𝗻𝗴 “𝗙𝗿𝗲𝗲-𝗧𝗵𝗶𝗻𝗸𝗲𝗿𝘀” 𝗔𝗴𝗮𝗶𝗻𝘀𝘁 𝗜𝘁𝘀𝗲𝗹𝗳 Why do people with low political trust sometimes believe a former Kremlin spokesperson more than a professional journalist or a European geopolitics expert on the war in Ukraine?

110

Antoine Marie PhD @a-marie-sci.bsky.social · 16/09/2026

𝐋𝐞𝐬 𝐋𝐋𝐌𝐬 𝐜𝐨𝐦𝐦𝐞 𝐚𝐭𝐨𝐮𝐭𝐬 𝐜𝐨𝐧𝐭𝐫𝐞 𝐥𝐚 𝐩𝐨𝐥𝐚𝐫𝐢𝐬𝐚𝐭𝐢𝐨𝐧 𝐞𝐭 𝐥𝐚 𝐦𝐞́𝐬𝐢𝐧𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧 Vous avez toutes et tous en tête les inquiétudes légitimes sur les risques que posent les LLMs.

122

Antoine Marie PhD @a-marie-sci.bsky.social · 17/07/2026

Mon apparition sur la chaîne Twitch de "Vous avez le droit" (aka @SébastienCanevet), le premier vulgarisateur du droit en France: m.twitch.tv/videos/28205...

110

Antoine Marie PhD @a-marie-sci.bsky.social · 06/07/2026

𝐅𝐚𝐮𝐭-𝐢𝐥 𝐬𝐞 𝐦é𝐟𝐢𝐞𝐫 𝐝𝐞𝐬 𝐡𝐨𝐦𝐦𝐞𝐬 𝐬𝐞 𝐝𝐢𝐬𝐚𝐧𝐭 𝐟é𝐦𝐢𝐧𝐢𝐬𝐭𝐞𝐬 ? / 𝐒𝐡𝐨𝐮𝐥𝐝 𝐖𝐞 𝐁𝐞 𝐖𝐚𝐫𝐲 𝐨𝐟 𝐌𝐞𝐧 𝐖𝐡𝐨 𝐂𝐚𝐥𝐥 𝐓𝐡𝐞𝐦𝐬𝐞𝐥𝐯𝐞𝐬 "𝐅𝐞𝐦𝐢𝐧𝐢𝐬𝐭𝐬"?

101

Antoine Marie PhD @a-marie-sci.bsky.social · 28/05/2026

les personnes les plus engagées moralement étaient aussi les plus susceptibles d'accepter cette conclusion erronée.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 28/05/2026

Leur conviction morale les rendait plus attentifs à des données solides — et c'est exactement ce qu'on espère.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 27/04/2026

People high in intellectual humility were consistently less likely to share hostile and false news. The takeaway: genuine epistemic rationality isn't about loudly moralizing logic. It requires something harder — restraining the ego, trusting your hunches less, and trusting others more.

141

Antoine Marie PhD @a-marie-sci.bsky.social · 27/04/2026

Across three large U.S. studies (N = 4,800, two preregistered, including a nationally representative YouGov sample), we find that people who strongly moralize rationality are more likely to share hostile outgroup news — including outright fabricated stories.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 10/04/2026

open.substack.com/pub/antoinem...

100

Antoine Marie PhD @a-marie-sci.bsky.social · 10/04/2026

New post on my Substack summarizing our ambitious cross-national study of online hostility, published in @nature.com Nature Human Behavior - with @boralexander.bsky.social @m-b-petersen.bsky.social @leapradella.bsky.social

162

Antoine Marie PhD @a-marie-sci.bsky.social · 25/03/2026

New post by me on @mchangama.bsky.social's substack The Bedrock Principle on “The psychology of speech repression”. Thanks Jacob Mchangama! Link below. 👇

142

Antoine Marie PhD @a-marie-sci.bsky.social · 21/03/2026

Robert Trivers, a theoretical genius and likely the Einstein of evolutionary biology, passed away. His theories of reciprocal altruism, parental investment, and self deception are among the deepest and most encompassing in the social sciences. Great post on them by @lionelpage.bsky.social

571

Antoine Marie PhD @a-marie-sci.bsky.social · 20/03/2026

Our paper just hit 50 citations 🎯: "Political conspiracy theories as tools for mobilization and signaling", Current Opinion in Psychology, 2022, with @m-b-petersen.bsky.social www.sciencedirect.com/science/arti...

196

Antoine Marie PhD @a-marie-sci.bsky.social · 05/03/2026

La présence de l’IA générative dans la couverture du conflit en Iran, ou la campagne des municipales françaises, réactualisent l’idée que nous serions parvenus à une époque de “post-vérité”.

110

Antoine Marie PhD @a-marie-sci.bsky.social · 18/02/2026

Mort de Quentin D. à Lyon : mon nouvel article dans The Conversation sur la radicalisation politique, co-écrit avec Peter Barrett. En accès libre ci-dessous. L'un des concepts clés est celui de "l'escalier de la radicalisation" : progressif et insensible, mais pouvant aboutir à l'irréparable.

101

Antoine Marie PhD @a-marie-sci.bsky.social · 07/02/2026

La mort d'Alex Pretti, abattu par des agents fédéraux à Minneapolis le 24 janvier dernier, était d’une barbarie inouïe pour un pays démocratique.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 22/12/2025

🚀 Le 24 décembre de 10h à 11h, je parlerai sur France Inter de comment rendre nos discussions politiques en famille plus productives et moins conflictuelles. Ce sera lors de l'émission d'Eva Roque, "SOS Noël" (lol), avec l'humoriste Florent Peyre. www.radiofrance.fr/franceinter/...

110

Antoine Marie PhD @a-marie-sci.bsky.social · 25/11/2025

New blog post on why it's not the best idea to talk about all Rassemblement national voters as a homogenous bunch of "racists". In English and French below.

101

Antoine Marie PhD @a-marie-sci.bsky.social · 19/11/2025

Lay people don’t response easily to large quantitative contrasts in policy efficiency when described in the abstract. Further, people seem even less sensitive to policy efficiency when they *moralise* the issue highly: they'll support almost any policy scenarios aiming to reduce CO2 emissions...

100

Antoine Marie PhD @a-marie-sci.bsky.social · 19/11/2025

high efficiency policies at least 40 times more impactful than low efficiency policies--and hugely profitable by a hundred million euros rather than costing this amount--received only a bit more support than low efficiency policies.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 19/11/2025

We presented naïve French respondents to policy scenarios in which a CEO or minister deployed a public policy meant to reduce CO2 emissions through carbon capture and storage technology.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 19/11/2025

🔥Laypeople Have Difficulty Processing Efficiency When Assessing Environmental Policies New Open Access paper on people's judgments of environmental policies with @hugo_trad and @stricklandbrent at @bppjournal.bsky.social Behavioral Public Policy 🔥 www.cambridge.org/core/journal...

123

Antoine Marie PhD @a-marie-sci.bsky.social · 15/10/2025

🔷 First, hyper-sensitive dispositions to detect threat, from social threats in particular. 🔷 Second, motivations to try to keep people *mobilized* for moral causes and against dangerous groups, by controlling information flows and beliefs. Basically, a policing motivation.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 15/10/2025

🔥 All political movements, left and right, engage in speech repression when it favours their interests and worldviews, or offers status-enhancement opportunities. Even those claiming to be “pro-freedom”. The default is a state of *moral hypocrisy*.

192

Antoine Marie PhD @a-marie-sci.bsky.social · 15/09/2025

With the same unbelievable courage, Daryl Davis has approached, befriended, and caused tens of KKK members to leave the klan. Here's an awesome interview in which he details his approach with Yascha Mounk: www.youtube.com/watch?v=hvt5...

110

Antoine Marie PhD @a-marie-sci.bsky.social · 15/09/2025

With breath-taking courage, Deeyah approached the racists in their homes to film them. By showing them respect and empathy, and asking the right questions, she gets them to feel dissonance, to realise their hateful generalisations make no sense. To watch urgently if you haven't.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 15/09/2025

Deeyah Khan just released on YouTube her magnificent documentary on the US "neonazis" (Meeting the enemy, 2017). It is one of the most moving and important documents there is out there about political hatred and how to reduce it: www.youtube.com/watch?v=hvt5...

141

Antoine Marie PhD @a-marie-sci.bsky.social · 08/09/2025

English translation of my essay expressing skepticism of the notion of a "post truth" era--from a cognitive perspective. Critical comments welcome.

101

Antoine Marie PhD @a-marie-sci.bsky.social · 05/09/2025

*Why is it taboo to publicly express discriminatory or "dangerous" statements? A "common knowledge" approach.* In all human societies, there exist "taboo" statements and topics in the sense that their public expression causes fear or is strongly morally disapproved of.

110

Antoine Marie PhD @a-marie-sci.bsky.social · 01/08/2025

I seem to see what I’d like to call the “jihadism-in-Gaza” fallacy in how some informed people discuss the conflict. What do you think?

110

Antoine Marie PhD @a-marie-sci.bsky.social · 15/07/2025

3) We provide a user-friendly website for annotating your texts and running prompt optimisation (no coding required)

000

Antoine Marie PhD @a-marie-sci.bsky.social · 15/07/2025

We 1) show that prompt choice with ChatGPT significantly affects performance at text classification tasks for social science, 2) We present a method for *automatically* optimising prompts, and

100

Antoine Marie PhD @a-marie-sci.bsky.social · 15/07/2025

New open access paper in the Journal of Computational Social Science with mathematicians Louis Abraham and Charles Arnal: "Prompt Selection Matters: Enhancing Text Annotations for Social Sciences with Large Language Models" doi.org/10.1007/s420...

131

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

Study 3 also shows that although social motivations for sharing the news (being liked, getting a news item fat checked) are not dominant compared to perceived accuracy and moral convictions, they are more at play when the political ingroup is watching:

110

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

Dealing with an ingroup (vs. outgroup) audience systematically increased intentions to share news congruent to the user and her audience (i.e., it amplified partisan sharing).

100

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

Study 1 provides a first external validity shot. Using data on news domains shared on the US Twitter in 2019, it shows that Twitter users engage in partisan sharing only when, and all the more as, they perceive their (real) followers as being somewhat to definitely likeminded.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

2) Users’ tendency to prefer sharing stories that promote their political convictions and partisan affiliations, even if such news fuels hostility between groups or is fabricated or cherry-picking.

100

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

1) The existence of online networks of politically like-minded users, sometimes suspected of furthering belief polarization via “echo chamber” effects;

110

Antoine Marie PhD @a-marie-sci.bsky.social · 03/07/2025

🔥New Open access paper in @pnasnexus.org with @m-b-petersen.bsky.social 🔥 "Motivations to connect with like-minded audiences increase partisan sharing on social media" academic.oup.com/pnasnexus/ar...

1124

Antoine Marie PhD @a-marie-sci.bsky.social · 13/05/2025

> Second, motivations to try to keep people *mobilized* for moral causes and against dangerous groups, by controlling information flows and beliefs. Basically, a policing motivation.

120

Antoine Marie PhD @a-marie-sci.bsky.social · 13/05/2025

🔥New preprint on why political activists sometimes try to censor free speech🔥 “Speech repression and threat narratives in politics: social goals and cognitive foundations” osf.io/preprints/os...

1114

Antoine Marie PhD @a-marie-sci.bsky.social · 24/04/2025

💡Nouvel article dans Analyse Opinion Critique dans lequel je propose une analyse assez... critique de l'idée que nous serions parvenus à une époque de "post-vérité" Open access: drive.google.com/file/d/1Xaz3... AOC: aoc.media/analyse/2025...

121

Antoine Marie PhD @a-marie-sci.bsky.social · 15/03/2025

Currently reading History of free speech by @mchangama.bsky.social Speech repression is practiced by *all* political groups, right and left, when it favors their interests. It aims to preserve people’s commitment to moral causes, and signal devotion to political camps.

172

Antoine Marie PhD @a-marie-sci.bsky.social · 11/03/2025

3) provide a user-friendly website for annotating your texts and running prompt optimisation (no coding required)

100

Antoine Marie PhD @a-marie-sci.bsky.social · 11/03/2025

We 1) show that prompt choice with ChatGPT significantly affects performance at text classification tasks for social science, 2) present a method for *automatically* optimising prompts, and

100

Antoine Marie PhD @a-marie-sci.bsky.social · 11/03/2025

Happy to have assisted data scientists Louis Abraham and Charles Arnal on this working paper: "Prompt Selection Matters: Enhancing Text Annotations for Social Sciences with Large Language Models" arxiv.org/abs/2407.10645

130

Antoine Marie PhD @a-marie-sci.bsky.social · 26/02/2025

Une analyse cognitive et sociale en accès libre 👇: drive.google.com/file/d/1SARX... Lien l'Express: www.lexpress.fr/idees-et-deb...

100