If software gets solved, could an automated AI factory self-train robotics models?
JulesJacobs5 · x · 2026-08-29
In this thread on RL and the LLM S-curve, the post raises the key follow-up: if the current path holds, does that imply an automated AI factory could be built to autonomously train models for RL-able tasks like robotics? The reply argues software is obviously going to be solved, but that optimism lives inside the RL-environment box—outside software, computer use and math, the trajectory looks much weaker.
More from AGI Musings
- Agents Don't Hate the Worst Systems, They Need Them — matt_slotnick · 2026-08-29
- Opinion: AGI is Becoming Less Interesting as a Milestone; The Economy Won't Wait for Definitions — VraserX · 2026-08-29
- Zephyr: Real AI Impact Will Stem from Productivity Gains, Not Just Infra Buildout — zephyr_z9 · 2026-08-29
- AI governance researcher: OpenAI and Anthropic should publish loss-of-control evidence first — sjgadler · 2026-08-29
- LLMs are too helpful: Should we include child language in training data? — Heavy_Carpenter3824 · 2026-08-29
- Investor Spotlights Four AI Frontiers: Personal Models, Sensors, Sims, and Explorers — pzakin · 2026-08-29