We show that, like humans, LLM groups display a 'utilitarian boost' in moral dilemmas, endorsing more norm violations for the greater good compared to solo agents, but through different argumentation. This has exciting implications for AI alignment, multi-agent design & artificial moral reasoning.

Excited to share that my recent paper *many llms are more utilitarian than one * has been accepted to NeurIPS 2025! 🎉
#NeurIPS

Thanks to my amazing collaborators!
@razanbaltaji.bsky.social

Preprint: arxiv.org/abs/2507.00814

Hi Elisa. Very interesting. Do you think we can discuss this tomorrow during the break?

great workshop!

In beautiful Norrköping for #IC2S2 @ic2s2.bsky.social Tomorrow, I’ll be presenting our recent work: Collective Moral Reasoning in LLMs: Many LLMs are more utilitarian than one. Let's connect!

Perfect! See you :)

Norrkoping.

In Norrköping to attend
at #IC2S2 2025.

How do LLMs reason morally in groups? We found that multi-agent LLMs, like humans, show a utilitarian shift, but the reasons differ.

📍 Session: ABM July 22 11:00–12:30

#AI #LLMs #MoralAI #CollectiveCognition #MultiAgentSystems

Join us at #IC2S2 2025 in Norrköping!

How do LLMs reason morally in groups? We found that multi-agent LLMs, like humans, show a utilitarian shift, but the reasons differ.

📍 Session: ABM July 22 | 11:00–12:30

#AI #LLMs #MoralAI #CollectiveCognition #MultiAgentSystems

What an exciting day for cognitive science with a double feature @nature.com including tiny models (www.nature.com/articles/s41...) and less tiny ones (www.nature.com/articles/s41...).

New preprint: Do LLMs make different moral decisions when reasoning in collective?

We find that LLM collectives endorse welfare-maximizing actions more when in groups than solo runs, even at the cost of harming a minority.

📄 arxiv.org/abs/2507.00814

@razanbaltaji.bsky.social

I’ll present "Many Minds, Diverging Morals: Human Groups vs. AI in Moral Decision-Making," my recent work with Eric Schulz & RazanBaltaji, next week at SCIoI, Berlin.

Join us! Open to the public!

www.scienceofintelligence.de/event/anita-...

Excited to be at #ICLR2024 in Vienna to present our latest work on causal reasoning in humans and LLMs! 🧠🤖 We examined biases in causal judgments using Causal Bayesian Networks. Join us at the Re-Align workshop!
openreview.net/forum?id=544...