'Backprop died long ago': AI circles debate deep learning's coming overhaul
fkasummer · x · 2026-10-20
A viral exchange: one account claims deep learning will look "unrecognizable" within six months, with replacing backprop "just the tip of the iceberg"; another replies that "backprop was killed a long time ago." Speculative discourse about a coming paradigm shift in training methods, with no technical specifics.
More from Research
- EMNLP paper: diffusion LMs internally encode latent denoising timestep representations — gsarti_ · 2026-10-23
- KU Leuven and VIB open two Group Leader professorships in AI for Biology, start 2027 — jpillowtime · 2026-10-23
- Subspace Tuning curbs LLM catastrophic forgetting by freezing principal singular directions — scottniekum · 2026-10-23
- Perceptic launches PercepticOS, an AI workspace for 'scientific superintelligence' — nathanbenaich · 2026-10-23
- EPFL's Bosselut to talk long context as test-time learning at ELLIS seminar — andre_t_martins · 2026-10-23
- DRATS: Adaptive Task Sampling Boosts Worst-Task Performance in Multi-Task RL — Cohere · 2026-10-23