New preprint links intelligence to learnable novelty across machine learning and complex systems
drmichaellevin · x · 2026-07-22
- A new preprint, “Intelligence from Learnable Novelty,” argues that seemingly different notions of intelligence across statistics, machine learning, complex systems, and agent behavior can be unified by a single principle: the pursuit of learnable novelty.
- The authors derive a closed-form approximation of Epiplexity and claim it reveals a deep connection between Epiplexity and intelligence.
- By maximizing Epiplexity in different systems, they report interesting emergent behaviors, including complex soliton interactions in cellular automata and effects in image encoders.
- The thread suggests the paper offers a new lens for thinking about intelligence, novelty, and objective functions across disciplines.
More from Research
- New NBER paper on how organizations use AI completes a three-paper series — daveholtz · 2026-07-22
- OAT uses 100 successful trajectories to debug failing AI agents without failure labels — TheTuringPost · 2026-07-22
- MoE, the mixture-of-experts architecture behind many top LLMs — vista8 · 2026-07-22
- NexForge synthesizes agent training data from requirements and lifts Qwen3.5 by 30 points — nex-agi · 2026-07-22
- SeerGuard uses a world model to screen risky actions in mobile GUI agents — Xue Yu · 2026-07-22
- A 15-hour video fine-tune turns masked trajectories into a robot control interface — Hadi Alzayer · 2026-07-22