Research program
Agent societies
The question
Can a well-organised group of agents be more general than any one model inside it?
Today most AI systems are one model answering one request. We think general intelligence is more likely to come from a system: many agents with different skills, a shared memory, and clear rules for who does what.
Open problems
- How agents should split a large goal into parts and hand work to each other.
- How to stop errors from one agent spreading through the whole group.
- How a group decides it is finished, and how it proves the result is right.
Our predictions
Our view, with how sure we are. Each one links to the evidence it rests on.
On the road to AGI
AGI will arrive as a system, not a single model: models, tools, memory, verification and many cooperating agents working together.
What changes
A medicine whose target and molecule were found mainly with AI will be approved for patients by a major regulator.
On the road to AGI
Open standards for connecting agents to tools and to each other will become as common for agents as web standards are for websites.
What changes
AI systems will run whole cycles of research in narrow fields, from hypothesis to experiment design to analysis, with scientists approving each step.
Progress around the world
Public results from other teams that shape our view.
· Linux Foundation
Open standards for AI agents get a neutral home
The Linux Foundation formed the Agentic AI Foundation around the Model Context Protocol, goose and AGENTS.md, backed by major AI and cloud companies.
What it means for AGI: agents from different makers are getting shared ways to connect to tools and to each other, a precondition for large agent societies.
· Google DeepMind
AlphaEvolve: a coding agent that discovers better algorithms
An evolutionary agent built on Gemini models found a way to multiply 4×4 complex matrices with 48 multiplications, improving on Strassen's 1969 result, and improved on known answers for a share of over 50 open maths problems.
What it means for AGI: a loop of models proposing, testing and keeping the best ideas can produce new knowledge, an early form of self-improvement.
· Google researchers (arXiv)
Co-Scientist: a multi-agent system that proposes research hypotheses
Specialised agents generate, critique and rank hypotheses in a tournament. Drug-repurposing candidates it proposed for acute myeloid leukaemia were validated in lab (in vitro) experiments.
What it means for AGI: a team of agents, not a single model, did useful scientific work that was checked in a real lab.
Source: arxiv.org(opens external site)
· Microsoft Research and collaborators (arXiv)
AutoGen: multi-agent conversation for complex tasks
A framework where several agents with different roles talk to each other, use tools and involve people to solve tasks.
What it means for AGI: building with groups of agents has become a mainstream engineering pattern.
Source: arxiv.org(opens external site)
· Stanford University and Google researchers (arXiv)
Generative agents simulate a small society
Twenty-five agents with memory, reflection and planning lived in a simulated town and showed believable individual and group behaviour.
What it means for AGI: memory and reflection give agents continuity, and groups of agents produce behaviour no single prompt produces.
Source: arxiv.org(opens external site)
Publications
No publications yet.
Collaborations
Open to collaboration.