Something Big Is Happening

Shumer opens by comparing the current moment to early 2020 when most people dismissed Covid warnings. He argues the gap between public perception of AI and its actual capabilities is now enormous and dangerous, because it prevents people from preparing for massive disruption that is already underway in tech and spreading to all knowledge work.

Shumer describes how, as of February 2026, he can describe an app in plain English, walk away for four hours, and return to a finished product — no corrections needed. The AI writes tens of thousands of lines of code, opens the app, tests it by clicking through features like a human would, iterates on its own, and only presents the result when it meets its own standards. This is not a prediction; it's his current Monday workflow.

The GPT-5.3 Codex model released on February 5th, 2026 exhibited something that felt like genuine judgment and taste — an intuitive sense of the right call, not just the technically correct one. Shumer notes this was previously considered the exclusive domain of human intelligence, and establishes a rule of thumb: if a model shows even a hint of a capability today, the next generation will be genuinely good at it due to exponential improvement.

OpenAI's technical documentation for GPT-5.3 Codex states it was "instrumental in creating itself" — used to debug its own training, manage deployment, and diagnose evaluations. Anthropic CEO Dario Amodei confirms AI is writing "much of the code" at his company. This recursive self-improvement loop — where each generation helps build the next, smarter version — is what researchers call an intelligence explosion, and the people building it believe it has already started.