Lewis Hammond, Cooperative AI | Introduction to Cooperative AI

Aug 10, 2023

Lewis Hammond, a DPhil candidate at Oxford and acting executive director of the Cooperative AI Foundation, dives into cooperative AI's role in enhancing human-machine collaboration. He explains why cooperation is vital for tackling global issues and discusses the dual-use risks of cooperative AI capabilities. Delving into the difference between cooperative intelligence and alignment, Hammond highlights innovative approaches like Bayesian inverse planning and the AI Economist project for redistributive tax insights. He also outlines the mission of his foundation in advancing cooperative AI research.

Ask episode

AI Snips

Chapters

Books

Transcript

Episode notes

INSIGHT

Cooperation Drives Civilization-Scale Outcomes

Many of humanity's biggest problems are cooperation problems like climate change, war, and pandemics.
As societies become more powerful and connected, solving cooperation becomes both more necessary and more impactful.

INSIGHT

AI Amplifies Cooperation Risks And Opportunities

Advanced AI will affect cooperation by increasing actors' power and interconnectedness, which can both worsen and improve cooperative outcomes.
We should therefore study how AI can either exacerbate flash-crash style failures or enable better coordination.

INSIGHT

Aim For Differential Progress In Capabilities

Differential progress matters: we can aim to develop AI capabilities that enhance cooperation more than competition.
Identifying such capability trajectories (e.g., cheap-talk channels) may steer AI toward socially beneficial outcomes.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Lewis is based at the University of Oxford where he is a DPhil candidate in computer science and a DPhil affiliate at the Future of Humanity Institute. He is also the acting executive director of the Cooperative AI Foundation, an affiliate at the Centre for the Governance of AI, and a fellow at the Foresight Institute. His research concerns safety, control, and incentives in multi-agent systems and in practice spans game theory, formal methods, and machine learning. Before coming to Oxford he obtained a BSc in mathematics and philosophy from the University of Warwick and an MSc in artificial intelligence from the University of Edinburgh.

In this talk I will provide an introduction to cooperative AI – a growing subfield of AI (safety) research that seeks to improve the cooperative intelligence of artificial agents and thus their ability to increase the joint welfare of humans, machines, and groups thereof. After motivating the importance of cooperation in the context of AI, I will highlight the location of cooperative AI within the broader landscape of AI safety research (most of which considers only single-agent settings) and provide several examples of problems and progress in this area. I will also provide a brief introduction to the Cooperative AI Foundation, a new charitable organisation that seeks to build the field.

Session summary

The Foresight Institute is a research organization and non-profit that supports the beneficial development of high-impact technologies. Since our founding in 1987 on a vision of guiding powerful technologies, we have continued to evolve into a many-armed organization that focuses on several fields of science and technology that are too ambitious for legacy institutions to support.

Allison Duettmann is the president and CEO of Foresight Institute. She directs the Intelligent Cooperation, Molecular Machines, Biotech & Health Extension, Neurotech, and Space Programs, Fellowships, Prizes, and Tech Trees, and shares this work with the public. She founded Existentialhope.com, co-edited Superintelligence: Coordination & Strategy, co-authored Gaming the Future, and co-initiated The Longevity Prize.

Apply to Foresight’s virtual salons and in person workshops here!

We are entirely funded by your donations. If you enjoy what we do please consider donating through our donation page.

Visit our website for more content, or join us here:

Every word ever spoken on this podcast is now AI-searchable using Fathom.fm, a search engine for podcasts.

Hosted on Acast. See acast.com/privacy for more information.