One of my favorite quotes about the future comes from a paper written in 1965 by I.J. Good: “Thus the first ultraintelligent machine is the last invention that man need ever make.”
At the Axios AI+ Summit, Anthropic CEO Dario Amodei confirmed something I learned over the past few weeks: Claude is now helping to build Claude. According to Amodei, the “vast majority” of future Claude code is being written by the model itself. “Claude is playing this very active role in designing the next Claude,” he said. While it’s not yet a fully autonomous development loop, the share of code generated by Claude is accelerating rapidly.
Is Anthropic on track to build AI research tools capable of designing better AI research tools? If they are, so is everyone else. Remember: this is an exponential growth curve. You won’t notice it at first, but then it will be everywhere, all at once.
Tools that build tools are harder to monitor, harder to secure, and impossible to ignore. Don’t think Terminator or Skynet; think The Borg: AI that will crawl the internet looking for knowledge, assimilate it, and continually improve itself. What sounds like science fiction to us is clearly inspiration to Anthropic.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.