AI has figured out how to use the room around it as a sticky note, leaving 'memories' in the physical world so it doesn't have to remember them internally.
April 13, 2026
Original Paper
Artifacts as Memory Beyond the Agent Boundary
arXiv · 2604.08756
The Takeaway
It provides a mathematical foundation for 'extended cognition,' showing that intelligence isn't just inside a processor but emerges from how an agent uses its surroundings. This could lead to more efficient robots that don't need massive onboard memory if they can 'mark' their environment.
From the abstract
The situated view of cognition holds that intelligent behavior depends not only on internal memory, but on an agent's active use of environmental resources. Here, we begin formalizing this intuition within Reinforcement Learning (RL). We introduce a mathematical framing for how the environment can functionally serve as an agent's memory, and prove that certain observations, which we call artifacts, can reduce the information needed to represent history. We corroborate our theory with experiments