Long-Form Interview on Self-Improving Agents
Stefania_druga · x · 2026-07-17
An in-depth interview with Stefania Druga, Staff Research Scientist at Sakana AI, focusing on research agents and Recursive Self-Improvement (RSI).
Key topics include:
- Catastrophic forgetting during long-horizon agent research tasks, and how a "memory-centric" harness can mitigate it
- Sovereign AI vs. open-weights: building on controllable models rather than getting locked into a single vendor
- Why computer use remains a massive headache even as models grow more capable
- Domains better suited for self-improving research agents: math, genomics, rare diseases
- A discussion on which tasks should remain human, and what humans should do once automated away
More from AGI Musings
- AI is still not at a maturity plateau, the author argues — generativist · 2026-07-22
- Essay argues LLMs are externalized metacognition, not standalone intelligence — lnsip9reg · 2026-07-22
- A multipolar AI race will not automatically make AI go well, repost argues — JeffLadish · 2026-07-22
- Decentralized AI as the Antidote to Digital Feudalism in the Economic Singularity — srimisra · 2026-07-22
- Humanoid robot sorting packages in a warehouse sparks debate over job loss — MonaJalal_ · 2026-07-22
- You can outsource thinking, but not understanding, in the age of agents — Yuchenj_UW · 2026-07-22