Recursive Self-Improvement

Let's talk about recursive self-improvement. Before your eyes glaze over because of tech speak, it's essentially AI training itself to become smarter.

This idea began with the concept of the "intelligence explosion" in 1965 with British mathematician I.J. Good.

He described the concept that a machine could become smart enough to build a smarter machine, triggering a feedback loop that compounds toward superintelligence.

What is superintelligence? It's the hypothesis that AI's capabilities will one day surpass that of humanity. A scary thought indeed.

But as it always seems to be true with AI, there are two sides to this coin.

One side: AI is already accelerating its own development. Anthropic published an article stating that, of May 2026, more than 80% of the code merged into Anthropic’s codebase was authored by Claude.

The other side: a financially-motivated marketing narrative with a multi-billion-dollar incentive. Valuations depend on growth, so the more credible superintelligence sounds, the more money flows in.

As Anthropic and OpenAI prepare for their IPOs, the narrative that they are on a path to superintelligence is the most powerful growth story ever told. 💸