Can AI Learn Without Humans?

As AI training evolves from human feedback toward reinforcement learning and continuous improvement, the role of human expertise is changing.

But how close are we to building AI systems that can truly learn and improve on their own?

For this edition of New Defaults, Alex Atallah, Co-founder & CEO of OpenRouter, will take the stage alongside Garrett Lord, Co-founder & CEO of Handshake, to explore what it takes for AI to learn from experience, how close we are to building truly self-improving systems, and what this shift means for model development and the future of human work.

What we'll explore

• The evolution of AI training: From human feedback to reinforcement learning and continuous improvement.
• The role of human expertise: How specialized knowledge helps train models, and whether AI will eventually outgrow that dependency.
• The future of autonomous learning: What it takes for AI agents to learn from experience, evaluate their own performance, and improve independently.
• The future of work: How self-improving AI could reshape human roles, skills, and economic opportunities.

Join us for a thought-provoking conversation with founders, researchers, engineers, and AI leaders exploring the next frontier of AI development.

Can AI truly learn without humans? Let's find out.

Spots are intentionally limited.

Agenda

5:30-6:00 PM - Registration, Bites & Drinks

6:00-6:45 PM - Fireside Chat

6:45-7:00 PM - Q&A

7:00-8:30 PM - Networking, Bites & Drinks

***

OpenRouter is the unified interface for LLMs: 600+ models from 90+ providers, processing 700T+ monthly. Better prices, better uptime, no subscriptions.