Sabitlenmiş Tweet

🧠 Have you ever thought about how scientists, despite often attributing the human brain to unguided evolution via natural selection, draw inspiration from its intricate, hierarchical architecture for AI breakthroughs?
Sapient Intelligence's 27M-parameter Hierarchical Reasoning Model (HRM) mirrors the brain's processing, outperforming models like Claude 3.5 on ARC-AGI and Sudoku without vast data.
We achieve this by reverse-engineering the brain's efficient systems—as if they were purposefully structured—rather than simulating random processes.
This raises a logical point: If evolution alone explains such complexity, why does treating it as intelligently organized yield superior results? Perhaps examining assumptions about design in nature warrants deeper reflection.
English




















