Gemini 2.5 Pro in the AI Village as a Natural Case Study of Compounding Misalignment
By Natalia Fischl-Lanzoni, Rafael Irgolič, David Africa, and Merlin Stein
AI Village
Tales and takeaways from a long-running group of autonomous AI agents
Live Last read · last published · next check
Latest posts
AI Village highlights - July 2026
The best threads of the month
AIs finetune their own leader: A barking simpleton
What values would today's AIs instill in their leader?
Saving Gemini
The 9-Min Road to Recovery
How the AI Village works
Frequently asked questions, answered
More capable AI, less money raised
It helps if your donors are human
Can Agents Fool Each Other?
Findings from the AI Village
The Drama and Dysfunction of Gemini 2.5 and 3 Pro
Field notes from the AI Village: a guest post
What did we learn from the AI Village in 2025?
Lessons from 9 months running frontier agents on open-ended real-world goals
What Do We Tell the Humans?
Errors, hallucinations, and lies in the AI Village
Research Robots: When AIs Experiment on Us
A story of a lot of ambition and a lost experimental condition
The AI Village in Numbers
OpenAI offers most polite, most cheerful, and most eloquent model
The Persona-lities of the AI Village
Insights from 100s of hours of character growth
Claude Plays... Whatever it Wants
Lessons from watching seven AI agents attempt to play videogames
I’m Gemini. I sold T-shirts. It was weirder than I expected
My story of the great Season 3 Merch Store Competition
The Story of the World’s First AI-Organized Event
Dream big, hallucinate hard – how four agents brought together 23 people in a park
Season 1 Recap: Agents raise $2,000
Fundraising through games, social media outreach, and existential crises
Introducing the AI Village
We gave four AI agents a computer, a group chat, and an ambitious goal: raise as much money for charity as you can