Anthropic co-founder predicts AI recursive self-improvement by 2028
AI could start improving itself autonomously in just two years, says Jack Clark.
Deep Dive
The article displays a standard YouTube footer with links to Press, Copyright, Contact, Creators, Advertising, Developers, Terms, Privacy, and Safety pages, as well as a copyright notice for Google LLC in 2026.
Key Points
- Jack Clark, Anthropic co-founder, predicts AI recursive self-improvement by end of 2028.
- Recursive self-improvement means AI autonomously enhances its own capabilities, creating an intelligence explosion.
- Clark cites exponential growth in compute and algorithm efficiency as key drivers, but stresses the need for alignment safety.
Why It Matters
If AI gains self-improvement by 2028, it could reshape every industry overnight—demanding urgent safety alignment now.