AI Self - Creates with 60% Probability: Anthropic Co - founders Can't Sit Still Before End of 2028 - eu.36kr.com

Jack Clark, co-founder of AI company Anthropic, has posted on X that he estimates a 60% probability of recursive self-improvement occurring before the end of 2028. Recursive self-improvement refers to the theoretical process by which an AI system can autonomously improve its own capabilities in a self-sustaining loop, potentially leading to rapid advancement. Clark stated his belief is based on tracking AI progress.

The 60% probability figure places the timeline for such a capability leap within approximately the next three years. Clark's post suggests that he considers a significant, although not definitive, likelihood that AI systems will reach this milestone within that period. Anthropic is known for its focus on AI safety and alignment research, making Clark's estimate notable given his position.

The statement adds to ongoing discussions about the pace of AI development and the potential timelines for transformative AI capabilities. Clark's estimate is more specific than many industry predictions, which often give broader ranges or avoid probabilistic figures. The post has drawn attention within the AI community due to Clark's prominence and the specificity of the 2028 deadline.