korrents

← Training is no longer limited by data but by compute, because most of the data models learn from is now synthetic.

On the map

8 connected korrents · 8 moments on record from 1 Oct 2024 to 1 Sept 2026.

Same subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subject Training is no longerlimited by data but bycompute, because most of… JH Jensen Huang — holds since 2026-03-23 Scaling laws are still working, but the models people actually use run a few months behind the maximum capability Google can deliver, because the biggest model is too slow and expensive to serve. Scaling laws are stillworking, but the models… SP Sundar Pichai — holds since 2025-06-05 Large language models could still plateau, and that possibility should be held open even though no evidence of it has appeared. Large language modelscould still plateau… DH David Heinemeier Hansson — holds since 2026-08-26 Current AI struggles even to rediscover old laws of physics from data, and when it does, contamination from training is the first suspect. Current AI struggleseven to rediscover old… TT Terence Tao — holds since 2025-06-14 If nobody creates content any more because everyone just asks an AI, there is nothing left for AI to train on. If nobody createscontent any more… PL Pieter Levels — holds since 2026-08-08 The cost of compute is the only intrinsic limit on AI self-replication; every other constraint on agents will have to be built by humans. The cost of compute isthe only intrinsic… DB Dean W. Ball — holds since 2026-09-01 AI acceleration threatens 'less software': without the constraint of scarce hours, even 37signals will likely build too much. AI accelerationthreatens 'less… DH David Heinemeier Hansson — holds since 2026-07-26 Within a couple of decades there will not be many people who can write, because AI has removed the pressure to learn. Within a couple ofdecades there will not… PG Paul Graham — holds since 2024-10-01 Inference was never going to be the easy, cheap half of AI, because inference is thinking, and thinking is far harder than reading. Inference was nevergoing to be the easy… JH Jensen Huang — holds since 2026-03-23
same subject