The AI Takeoff Article Everyone Is Sharing

Gavin Purcell, writing about the AI "awareness gap":

The bubble I'm talking about is AI twitter and those of us who yammer back and forth about the next big AI feature all day long. Those of us who actually understand what's happening in the AI space and specifically... how AI takeoff is happening now and could leave a LOT of people behind.

There's a genuine observation buried in here - most working professionals aren't paying close attention to how quickly AI capabilities are advancing, and the gap between the AI-aware and AI-unaware is real. That part I agree with.

But then the piece does the thing that AI communication almost always does: it takes a reasonable observation and wraps it in the language of existential emergency. "Made my skin crawl." "Keeps me up at night." "We might never catch up."

The core argument rests on recursive self-improvement and an exponential curve illustration from a ten-year-old Wait But Why post. What it doesn't do is distinguish between a model contributing to its own training pipeline - which is what "working on itself" actually means in the GPT-5.3 Codex context - and the kind of unbounded intelligence explosion that the exponential curve implies. These are very different things. The piece treats them as the same thing because the conflation is what makes the narrative work.

Most telling is the advice section. After eight hundred words of building genuine alarm about an intelligence explosion that could "leave a LOT of people behind," the practical guidance is: be creative, invest in relationships, start a side hustle. If the exponential curve is as steep and imminent as described, "start a side hustle" is bringing a packed lunch to an asteroid impact.

The mismatch between the scale of the alarm and the modesty of the advice tells you everything. The author is less certain about the timeline than the tone suggests. Which is fine - honest uncertainty is the correct position. But honest uncertainty doesn't go viral. Skin-crawling does.

Send this to a professional who's anxious about AI and it will make them more anxious, not more competent. That's the test most AI communication fails right now.