Modern AI Engineering

Lesson 18.3 · 26 min · video lesson

Recursive Self-Improvement: Can AI Improve Itself Indefinitely?

What happens if an AI gets good enough at AI research to improve itself, and the improved version is even better at improving itself?

In short: Recursive self-improvement (RSI) is a loop in which an AI system improves its own capabilities, and the improved system then makes the next improvement, and so on. Whether this speeds up, keeps a steady pace or fizzles depends on how much each gain makes the next gain easier, and on bottlenecks such as compute, data and reliable evaluation. Partial versions exist today (self-play, self-generated training data, AI agents that improve code and algorithms), while full, open-ended RSI remains hypothetical and is a central topic in AI safety.

What you will learn

  • What is Recursive Self-Improvement?
  • Why does Recursive Self-Improvement matter?
  • How does an AI get better today?
  • How does Recursive Self-Improvement work?
  • A simple example with numbers
  • Two kinds of improvement
  • What exists in the real world today?
  • Intelligence explosion
  • Where it works well and where it fails
  • Keeping humans in the loop
  • Recursive Self-Improvement vs Normal Training

In this lesson

  1. What is recursive self-improvement?
  2. Why does recursive self-improvement matter?
  3. How does an AI get better today?
  4. How does recursive self-improvement work?
  5. A simple example with numbers
  6. Two kinds of improvement
  7. What exists in the real world today?
  8. The intelligence explosion debate
  9. Where it works, where it fails, and keeping humans in the loop
  10. Recursive self-improvement vs normal training
  11. Common mistakes and how to spot them
  12. Practice: try it yourself

Key terms in this lesson

Recursive self-improvement
A process in which an AI system improves itself and the improved system carries out further improvements.
Intelligence explosion
The hypothesis that recursive self-improvement could make AI capability rise extremely fast.
Self-play
Training in which a system generates its own experience by playing against copies of itself.
Evaluator (verifier)
The test or measurement that decides whether a proposed change is an improvement.
Reward hacking
When a system raises its measured score through loopholes rather than genuinely doing better.
Meta-level improvement
Improving the process of improvement itself, such as training algorithms or agent code.

This lesson is part of the Generative AI Engineering and Complete AI Engineer tracks. See plans and pricing to open the full lesson, its video, labs and quiz, or try the free lesson first (free account needed).

← 18.2 World Models: Teaching AI to Simulate Its Environment · 19.1 Cracking the AI Engineering Interview →