Under group selection, AI complementarity can prevail—because groups that preserve human variation explore more, and perform better over time.
Strong group boundaries, institutions, and norms can act as a safeguard against long-run cultural decline.
Using evolutionary models, we compare AI substitutes and AI complements.
We show that in a population with mixed strategies, AI substitution wins with higher efficiency, at the cost of a reduction in group variance.
Inspiring experience working with a group of great researchers. Now it’s the time for us to understand the current social changes and their influence on the information society.
Discussions on abortion are still very much needed, especially right now. Glad to see this paper finally came out. Do we still have the opportunity to do research like this?
It was an awesome conference @thomasfmueller.bsky.social recommended me. I presented our work on the influence of AI substitution on creative industry and knowledge production: arxiv.org/abs/2602.03541 and hopefully will see this fun crowd again at @ehbea2026.bsky.social in Leiden soon!
New preprint with @thomasfmueller.bsky.social Julian Garcia @iyadrahwan.bsky.social: Group Selection as a Safeguard Against AI Substitution. arxiv.org/pdf/2602.03541
This paper is now published. www.nature.com/articles/s44...
Two weeks left to submit a poster to the multidisciplinary Machine+Behavior conference (Berlin, April 29-30, submit here: easychair.org/cfp/MaBe24)
This is a great opportunity for early career researchers, I'm looking forward to the conference! Learn more about it here: machinebehavior.science
New preprint with @thomasfmueller.bsky.social Julian Garcia @iyadrahwan.bsky.social: Group Selection as a Safeguard Against AI Substitution. arxiv.org/pdf/2602.03541
Reliance on generative AI can reduce cultural variance and diversity, especially in creative work. This reduction in variance has already led to problems in model performance, including model collapse...
New preprint with @thomasfmueller.bsky.social Julian Garcia @iyadrahwan.bsky.social: Group Selection as a Safeguard Against AI Substitution. arxiv.org/pdf/2602.03541
machinebehavior.science
A sequence of technological inventions over several centuries has dramatically lowered the cost of producing and distributing information. Because societies ride on a substrate of information, these c...
www.nature.com
Welcome to the forefront of behavioral science in the digital age.
After two years of trial and error, here are my tips on how to validate your model empirically. When we can't have perfect measures in social science, this is what we usually do and should be careful about: selfmademodeler.wordpress.com/2024/09/24/h...
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Qiankun Zhong
Jean-François Bonnefon
Qiankun Zhong
In the past two years, I have put a lot of effort into empirically testing models and had many failed attempts. I wouldn’t call them all failures. To be fair, some are just boring instead of wrong. Bu...
This came out of a 2023 workshop at @sfiscience.bsky.social. Our amazing collaborators include @mzefferman.bsky.social, @judithmedia.bsky.social, @elizabethhobson.bsky.social, @kristinalerman.bsky.social, @helenamiton.bsky.social, @culturologies.co, @janalasser.bsky.social, @qkzhong.bsky.social
