Researcher interested in cultural evolution, AI, philosophy, also cinema & poetry
Stockholm, Sweden
Joined May 2009
- Tweets1.8K
- Following1.3K
- Followers677
- Likes7.1K
Pinned Tweet
As autonomous AI agents are deployed at scale, how can we ensure they cooperate with one another?
In our new paper, we find that selecting for group performance can effectively promote cooperation among LLM agents.
arxiv.org/pdf/2606.23343
Aron Vallinder retweeted
Under group selection, cooperation rose and stabilized. Under individual selection, self-interested prompts spread and the population moved toward defection.
Final-generation donations: 47.5% vs. 8.7%. No selection: 34.3%.
Some models also anticipated the selection pressure. When told how selection worked, GPT-5.4 changed its behavior before evolution began, opening a ~25-point cooperation gap. Llama shifted less; Qwen did not.
This was a joint effort with an incredible team: Luis Celiktemel, Edward Eichhorn, @LevinBrinkmann, @schimmelrob, @YaominJ , @edwardfhughes, and @iyadrahwan
The paper again: arxiv.org/pdf/2606.23343
More to come!
Aron Vallinder retweeted
If intelligence is inherently social, then the path to more powerful AI runs not through building a single colossal oracle but through composing richer social systems
Each prior “intelligence explosion” was not an upgrade to individual cognitive hardware, but the emergence of a new, socially aggregated unit of cognition. Primate intelligence scaled with social group size, not habitat difficulty.
Human language created what Michael Tomasello calls the “cultural ratchet”: knowledge accumulating across generations without any individual requirement to reconstruct the whole. Writing, law, and bureaucracy externalized social intelligence into infrastructure, institutions that coordinate across longer time horizons than any participant within them.
This article nails it: science.org/doi/10.1126/scie…
Aron Vallinder retweeted
Our new ‘Introduction to Cooperative AI’ curriculum is now available. Master the foundations of cooperative AI and gain confidence to contribute to the field. Link below.
Aron Vallinder retweeted
New paper at ICLR 2026! 🎉
"Language and Experience: A Computational Model of Social Learning in Complex Tasks"
We model how humans combine advice from others with direct experience to learn new tasks, and show this enables bidirectional human-AI knowledge transfer.
🧵⤵️
Jia Zhangke, one of the very best filmmakers working today, experiments with AI-generated video.
Certainly looks nowhere near as good as his other work, but I guess that's not the point either.
Aron Vallinder retweeted
Over 5 years I've advised dozens of philanthropists on AI. I compiled the answers to all of the questions I've been asked in one report.
2024 Nobel Prize Geoffrey Hinton calls it “an extremely useful resource for philanthropists interested in funding AI safety and preparedness."
Aron Vallinder retweeted
Excited to share our work on envisioning Intelligent AI Delegation (arxiv.org/abs/2602.11865). Delegation in most existing AI systems is brittle, and relies on simplified hand-crafted control flows. As such, it fails to meet the requirements of what is needed to truly scale distributed reasoning and task completion in multi-agent systems and collectives.
Aron Vallinder retweeted
1/ What makes self-interested AI agents cooperate? Not fine-tuning. Not central oversight. Gossip.
RIP Béla Tarr, one of the great directors.
I met him once at a film festival in 2008, and his first words to me were "Fuck off, guy!"
Aron Vallinder retweeted
I'm starting a newsletter in the spirit of my tweets: The Update.
theupdatebrief.substack.com/…
I'm aiming to distil the most important news, analyses, and Twitter debates for you.
Tell your friends, and please subscribe!
Aron Vallinder retweeted
New RFP! We give out $5M in grants annually to reduce existential risks from advanced AI. Now we are seeking projects across 39 research areas within security, coordination, research automation, and neurotech.
More about the specific work we are interested in funding, and why: foresight.org/request-for-pr…
If you know people doing great work in these fields – please let them know! The next application deadline is June 30th.