Evolving AI — ProSocial World

Cambridge, Massachusetts · October 8–10, 2026 · Invitation only

Evolving AI

A gathering to catalyze modern evolutionary thinking in the AI community.

Two communities — artificial intelligence and evolutionary science — will meet at Harvard to think together about what AI means as an evolutionary event, and what evolutionary science has to offer a field that hasn't yet made much use of it.

The premise

A turning point, viewed through an evolutionary lens

Artificial intelligence is a turning point in human history. It can even be described as a major evolutionary transition (MET), comparable to previous METs in human evolution, such as the emergence of spoken language, and in biological evolution, such as the evolution of multicellular organisms and ultrasocial animal societies. Yet, modern evolutionary science, including knowledge about these other METs, is not yet being leveraged in the field of AI.

There is a reason for this vacuum of knowledge. Evolutionary science was confined to the study of genetic evolution for most of the 20th century. The study of other evolutionary processes, such as human cultural evolution and evolutionary algorithms on computers, is more recent and only slightly predates the AI revolution.

MET — Major Evolutionary Transition

The framework connecting the emergence of language, multicellularity, and ultrasociality — and now situating artificial intelligence as the newest such transition in the history of life on earth.

The gathering

Fifty minds, held in catalytic proximity

Catalysis here is more than a poetic metaphor: catalytic molecules grab onto other molecules, hold them in an orientation that binds them to each other, and are released to repeat the process. The reaction proceeds orders of magnitude faster than it otherwise would.

This invitation-only gathering will act as a form of cultural catalysis.

25

The AI community

Members who appreciate the need for an evolutionary perspective on the technology they are building.

25

Evolutionary science

Members who have already begun thinking about, and working in, artificial intelligence.

Themes

Five threads, one evolving conversation

The topics and questions we'll explore — where evolutionary theory meets the practice of building intelligent systems.

01

Human–AI cooperation

How can humans and intelligent systems coordinate toward socially beneficial outcomes?

02

Evolution of agentic systems

What dynamics — cooperation, competition, organization — emerge among interacting AI agents?

03

Open-ended learning

How can AI continuously generate novel, useful capabilities without fixed objectives?

04

Biological and machine learning

Which capabilities of biological learning are missing from current AI, and how might AI expand our own?

05

Human adaptation to superintelligence

How can human expertise, judgment, and agency stay valuable as AI capabilities advance?

Framework

Tinbergen's four questions

Four complementary bodies of knowledge, together needed for a full understanding of anything evolution has produced — a birdsong, an institution, a language model.

Function

Why?

The adaptive purpose a trait serves — what it is for.

History

Whence?

The lineage behind it — how it came to be, over time.

Mechanism

How?

The immediate machinery that makes it work in the present.

Development

From what?

How it assembles and unfolds over a single lifetime.

Evolving AI Fellowships

For the next generation of students, early career researchers, and builders

We are pleased to provide fellowships for the next generation of students, early career researchers, and builders to join this gathering and be part of the first Evolving AI Fellowship cohort.

Essay question

"What do you think the future of human-AI interaction will look like in 5 years? What role do you think evolutionary science has to play in that future?"

01

Basic information

Basic demographic information

02

Current CV

Uploaded as part of the application

03

Essay response

Under 500 words, answering the question above

Applying as a group?

Applications can be submitted as either an individual or a small team of 2 or 3 individuals who have a history of working together on research, building, or both. If you are submitting an application as a group, you will each have to submit a separate application and include a brief explanation for why you are applying collectively, a description of your history working together, and what you hope to accomplish through attending as a group.