Anthropic's Exponential AI Growth Claim Faces Skepticism From Inside and Outside

Anthropic CEO Dario Amodei argues AI progress will follow exponential scaling laws, suggesting we could reach powerful AI if current computing-power trends continue. But Anthropic’s own model cards for Claude Mythos and Fable show no clear AI-attributable acceleration, and critics like Gary Marcus have long questioned scaling laws, hinting at possible diminishing returns. The piece frames the debate as ongoing: is AI progress truly accelerating, or hitting limits, with references to OpenAI scaling work and GPT-3’s observed shortcomings.