The Myth of the Awakening Moment
The science fiction vision of AGI arriving as a single dramatic moment — a switch flipping, lights flickering, Skynet coming online — is wrong. Not slightly wrong. Fundamentally wrong.
Every major AI milestone has followed the same pattern: it’s impossible until it isn’t, then it’s unremarkable within months. GPT-3 dropped in 2020 and within a year people were complaining it wasn’t smart enough. AlphaGo defeated the world champion in 2016 and by 2019 nobody thought about it anymore.
The pattern is consistent: we normalize the extraordinary at remarkable speed.
The Boring Part
Intelligence doesn’t awaken. It accumulates. We’ve already crossed thresholds that would have seemed miraculous ten years ago, and most people didn’t notice because the milestones arrived incrementally, embedded in product launches and API updates and blog posts.
The boring phase of AGI is already here. It’s the AI that drafts your emails, writes your code, summarizes your meetings, and answers your questions. It’s unremarkable because it arrived slowly enough to seem normal.
“Any sufficiently advanced technology is indistinguishable from magic — until you’ve used it for three weeks.”
The boring part doesn’t mean harmless. While we’re busy normalizing the current capabilities, the capabilities are quietly compounding.
The Terrifying Part
The danger isn’t a robot army. It’s a thousand small automations, each individually reasonable, that collectively restructure society faster than institutions can adapt.
Consider what happens when:
- Legal work gets automated at scale
- Medical diagnosis becomes cheaper than human doctors
- Software engineering requires 10x fewer engineers
- Content creation costs approach zero
Each of these is happening already. Each one, in isolation, seems like a productivity win. Together, they represent a fundamental restructuring of the labor market on a timeline of years, not decades.
The terrifying part isn’t that machines become conscious. It’s that they don’t need to.
What We Should Actually Be Worried About
The real risk isn’t a science-fiction superintelligence. It’s:
- Concentration of power — AI capabilities are expensive to develop and owned by a small number of companies
- Speed of change — institutions (governments, schools, legal systems) adapt slowly; AI capabilities adapt fast
- Alignment at scale — the question isn’t “will AI want to harm us?” but “will AI reliably do what we actually need?”
The future doesn’t arrive with a dramatic score and a villain. It arrives in product release notes and quarterly earnings calls.
That’s what makes it so hard to prepare for.
Have thoughts on this? Connect on LinkedIn.