Home / Technology / Can AI Build Its Own Successor by 2028?
Can AI Build Its Own Successor by 2028?
4 Jul
Summary
- AI may achieve recursive self-improvement by 2028.
- Current AI models can already build smaller models.
- Concerns rise over loss of human control in AI development.
Artificial intelligence development is advancing rapidly, with predictions that AI systems could create their own successors by 2028. This phenomenon, known as recursive self-improvement (RSI), involves AI models building increasingly capable versions of themselves without human intervention. While current AI can assist in building smaller models, full automation of the AI development process is becoming a near-term possibility.
This accelerating progress has sparked debate and concern among researchers. Some worry that RSI could lead to superintelligence beyond human control, with potential consequences ranging from loss of human decision-making power to existential risks. Experts suggest that current AI models are already capable of tasks previously requiring human coders and engineers, and this trend is expected to continue.
However, physical constraints like compute access and the availability of training data may currently limit the speed of AI self-improvement. The demand for AI services also competes for resources needed for AI research and development. Despite these limitations, the potential for AI to significantly boost research and development productivity is undeniable, prompting calls for cautious management of this technological frontier.
