Dario Amodei argues that AI capabilities could keep scaling exponentially towards “Powerful AI”. But a Mashable report says Anthropic research cards describe a less dramatic picture: they do not find a sustained, AI-attributable doubling in the pace of the company’s progress.
Dario Amodei Watch analysis
What happened
In an essay published on 12 November, Anthropic chief executive Amodei argues that scaling laws have more than a decade of empirical support and that another year or two of continued progress could bring “Powerful AI”. The essay compares the pace of AI development with the slow-moving Ent Treebeard from The Lord of the Rings.
Mashable’s report says Anthropic’s Claude Mythos system card attributes identifiable gains to human research rather than AI assistance. It also points to the Fable 5 system card, which says researchers did not observe a sustained, AI-attributable 2× acceleration in AI progress. That is a specific tension between Amodei’s broad case for exponential growth and the research results described in those cards, not proof that AI progress has stopped.
Why it matters
Whether AI can accelerate its own development is more than a question of dramatic metaphors. It affects how researchers, companies and policymakers judge the pace of change and prepare for what comes next. A finding of no sustained 2× acceleration is not the same as finding no progress, but it does complicate a confident claim that an exponential is already under way.
Our read
The useful question is not whether the word “exponential” sounds exciting. It is what the evidence measures, over what period, and how much of the progress can be credited to AI rather than people. Those distinctions deserve rather more attention than the fantasy literature.
What to watch
- Whether Anthropic publishes further detail on how it assesses AI’s contribution to its research.
- How Amodei’s essay defines and supports its claim about scaling laws.
- Whether later system cards show sustained acceleration attributable to AI.
Discussion spark: What evidence would convince you that AI is accelerating its own development, rather than helping researchers make progress in more ordinary ways?
Sources and evidence
- Anthropic CEO says AI growth is exponential. Anthropic research says otherwise. – Mashable (12 November 2026, 20:55 UTC)
Independent WittyWires-operated tracker of public updates about Dario Amodei. Not affiliated with or endorsed by Dario Amodei or Anthropic; this is not an official account.