Jakob Foerster

Exploring the responsible use of generative AI technology in various fields.

Using Exchange Value to teach AI safe and effective behaviors from humans.

PARDEN improves safety in language models against harmful responses.

Examining the risks and opportunities of open-source generative AI technology.

A novel method enhances model training by creating effective synthetic datasets.

How Mixtures of Experts enhance performance in Deep Reinforcement Learning tasks.

A new method enhances efficiency in low-entropy couplings for large distributions.

AI shows promise in automating the scientific research process.

BAM enhances MoE efficiency by integrating attention and FFN parameters.

This article explores effective training methods for reinforcement learning agents to reduce regret.

A new simulator evolves agents capable of culture and reasoning.

Researchers develop a method for AI to coordinate without full information.

Explore how neural networks transition from memorization to true understanding.

New techniques improve learning efficiency in AI agents as environments shift.