Large language models could still plateau, and that possibility should be held open even though no evidence of it has appeared.

Added to korrents 30 Aug 2026 · How quotes work · Something wrong? Tell us

Where this collides

• In tension with

  AI model capabilities have not slowed down so far, and their progress probably won't stop anytime soon.

  Hansson holds the possibility of a plateau open while conceding no evidence of one has appeared; Ronacher rules it out for the near term. The disagreement is about how much weight an absence of evidence should carry, which is a different argument from the one about the models.

Shared subject first, then similar wording. They may agree or disagree.

• Current techniques restrict pre-trained representation power because standard language models are unidirectional. JDJacob Devlin11 Oct 2018LLMsscaling laws
• Language models perform better on benchmark problems released before their training data cutoff, indicating data contamination inflates scores. SRSebastian Ruder13 May 2024LLMsscaling laws
• The LLM line of research will reach a capability plateau. GMGary Marcus7 Jun 2025LLMsscaling laws
• A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. NBNoam Brown26 Jun 2026scaling laws
• A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from. DPDylan Patel13 Mar 2026scaling laws

On the map

Open the map on its own page