Ten gems from Diogo's Code Summit talk: RLHF is part of the problem, software is back
TheTuringPost · x · 2026-11-12
Turing Post compiled the top-10 takeaways from TypeSafe AI's Diogo (@CompleteSkeptic) talk at AI Engineer Code Summit:
- "We've been announcing AGI three months away for two years," yet the world outside SF looks unchanged
- Models are "Einstein 95% of the time, Mr. Bean 5%" — you can't automate what you can't predict
- "RLHF is part of the problem": we optimized for assistance with human review, then expected autonomous reliability
- "The bitterest lesson is doing the right task": task choice beats data, which beats compute
- The world has "LLM Stockholm Syndrome": we design around today's limitations and still ship the same software with a chatbot bolted on
- Put intelligence into the software's inner loop; make intent understanding a reliable building block
- Real engineering: break tasks down, build verifiable stages, manage state — stop every agent rediscovering the same patterns
- Slop thrives because bad actors enjoy an asymmetric advantage: readers pay the cost of filtering
- Do "software research": build weird things with no business case; 1% may be awesome
- "Software is gonna be so fing back": the question shifts from "can we automate this?" to "how much work will it take?"
More from AGI Musings
- AI-cures thesis hinges on robotics and autonomous labs generating data first — shae_mcl · 2026-11-12
- CrewAI Founder: Anti-American-AI FUD Won't Stop Until Volatility Shakes You Out — robleclerc · 2026-11-12
- Kai-Fu Lee wraps 12-hour London blitz promoting 'AI Native' book launch — kaifulee · 2026-11-12
- Do closed AI labs' research agendas diverge? Talent flow acts as a penalty term — stochasticchasm · 2026-11-12
- AI 'doomers' call themselves 'the Cassandra fringe' — critic says the analogy flatters overconfidence — DrTechlash · 2026-11-12
- AI agent traffic reportedly grew 8x in just two months — Dr_Singularity · 2026-11-12