Generalising from Self-Produced Data: Model Training Beyond Human Constraints
This proposes a foundational shift toward autonomous general intelligence, addressing the problem of human-imposed limitations in AI training for the broader AI research community.
The paper tackles the limitation of LLMs being constrained by human-derived data and introduces a framework where AI models autonomously generate and validate knowledge through environmental interaction to maximize an unbounded numeric reward, aiming for self-improving systems beyond human constraints.
Current large language models (LLMs) are constrained by human-derived training data and limited by a single level of abstraction that impedes definitive truth judgments. This paper introduces a novel framework in which AI models autonomously generate and validate new knowledge through direct interaction with their environment. Central to this approach is an unbounded, ungamable numeric reward - such as annexed disk space or follower count - that guides learning without requiring human benchmarks. AI agents iteratively generate strategies and executable code to maximize this metric, with successful outcomes forming the basis for self-retraining and incremental generalisation. To mitigate model collapse and the warm start problem, the framework emphasizes empirical validation over textual similarity and supports fine-tuning via GRPO. The system architecture employs modular agents for environment analysis, strategy generation, and code synthesis, enabling scalable experimentation. This work outlines a pathway toward self-improving AI systems capable of advancing beyond human-imposed constraints toward autonomous general intelligence.