As they say, every exponential is a sigmoid in disguise. I think the exponential phase of growth for LLM architectures is drawing to a close, and fundamentally

As they say, every exponential is a sigmoid in disguise. I think the exponential phase of growth for LLM architectures is drawing to a close, and fundamentally new architectures will be necessary for meaningful advances. I'm also not convinced by the graphs in this article. OpenAI is notoriously deceptive with their graphs, and as Gary Marcus has already noted, that METR study comes with a lot of caveats: [https://garymarcus.substack.com/p/the-latest-ai-scaling-graph-and-why?utm_source=publication-search https://garymarcus.substack.com/p/the-latest-ai-scaling-grap...]