Deep learning is 'brute force mixed with spray & pray': weights can't encode the geometry of epistemics
ryunuck · x · 2026-09-21
Extending his critique of RL environments, the author argues that because the substrate is activation functions and weights, knowledge is distributed across the network with no localized home — the model is geometrically unstructured with respect to the geometry of epistemics, so holes can't be causally targeted. His verdict: deep learning is brute force mixed with spray & pray.
Related event: Debate: Can LLMs Have Real Cognitive Structures?(2 posts)→
More from Research
- Block-sparse featurizers show visual concepts live on 2-4D manifolds, says 26-author arXiv paper — burny_tech · 2026-09-21
- ICLR review anonymity debate: most conferences still let senior reviewers and PCs see real names — AnkaReuel · 2026-09-21
- dair-ai weekly: 7 top AI papers led by Byte Model Scaling — dair_ai · 2026-09-21
- Researchers publicly call out senior academics for rushing out 'blatant slop' papers — thegautamkamath · 2026-09-21
- Founder says frontier lab's Jev copies his 2025 non-autoregressive decision model paper and open weights — abhijithneil · 2026-09-21
- Schmidhuber: LeCun's JEPA is essentially his 1992 Predictability Maximization system — SchmidhuberAI · 2026-09-21