The Other AI
We've settled on the wrong villain.
The dominant story about artificial intelligence goes like this: a tireless, inhuman force is coming for the things that make us us — our jobs, our judgment, eventually our relevance. It's a good story. It has a clear antagonist and a satisfying sense of dread. It is also, I think, the least interesting way to understand what's actually happening.
Here's a reframe worth sitting with: the most important collaboration of the next decade isn't AI versus us. It's AI plus the other AI: actual intelligence. All credit to Steve Wozniak for the clever reframing, which he shared during his commencement address to Grand Valley State University on May 1, 2026. The line got a standing ovation, and I don't think that was an accident. In a season of commencement speeches where audiences booed anyone who cast AI as the protagonist of their future, Wozniak got cheered for putting the humans back at the center of the sentence. That's the whole thesis, in five words from a man who helped create the defining machines of our era.
"You have AI — actual intelligence." Steve Wozniak, commencement address, Grand Valley State University, May 1, 2026
Actual intelligence, in Wozniak's sense, is the messy, intuitive, contradiction-tolerant kind that has always lived in human skulls. Pair it deliberately and well with its artificial cousin, and you don't get replacement. You get a kind of creativity neither one can produce alone.
I think about this in terms of what each side is actually good at, because the two AIs are good at almost entirely different things. Artificial intelligence is extraordinary at search — at trying ten thousand variations of a molecule, a material, a configuration, and finding the one that works. What it is not good at is knowing which question was worth asking in the first place. That's a taste problem, a judgment problem, a "does this actually matter" problem. It's the oldest human faculty there is, and it hasn't been automated, because it isn't really a computation. It's a bet, informed by everything a person has lived and read and gotten burned by.
I spent years covering medical technology as a reporter, watching this exact tension play out long before anyone called it AI. A diagnostic algorithm can flag an anomaly in a scan or a sensor stream faster and more consistently than any clinician alive — that part is no longer in dispute. What it can't do is decide whether that anomaly is a signal worth building a device around, whether the defect it just caught is actually dangerous or just looks scary in a report, or whether a regulator will read the data the same way the engineers do. I sat through enough FDA advisory panels to know that the hardest calls in medtech were never the ones with a clean data answer. They were the ones where the data said one thing and twenty years of clinical judgment said "wait." That's actual intelligence doing what it has always done: making meaning out of noise, and knowing which noise to trust. The machine finds the anomaly. The human decides whether it's a discovery, a false alarm, or a reason to pull a product before it ever reaches a patient.
The mistake we keep making is treating this as a tug-of-war for the same rope — as if every task artificial intelligence gets better at is a task human intelligence has lost. That's not how good pairings work. A telescope didn't diminish the astronomer; it changed what an astronomer's attention was for. The best scientists I've spent time around don't experience their tools as competitors. They experience a good tool as an argument-starter — something that hands them a strange result and dares them to explain it.
I'd go further: paired well with AI, actual intelligence stands to expand human creativity and discovery further than we can currently picture. But that expansion doesn't shrink our job — it raises the stakes on it. We'll still need to lean on instinct, judgment, and hard-won wisdom to weigh the output from this unrelenting deluge of creativity to decipher which ideas are actually worth our attention and action. That recognition is not a smaller job than invention. It might be the harder one. It's certainly the more human one.
So I'd like to retire the villain. Not because the risks of artificial intelligence aren't real — they are, and they deserve rigorous debate about how we move forward responsibly. But the story we tell shapes the systems we build, and a story about replacement produces defensive systems, deployed reluctantly, by people bracing for loss. A story about pairing produces something else entirely: teams that treat the machine as an instrument for asking better questions, faster, so the humans in the room can spend their finite, irreplaceable judgment on the questions that actually deserve it.
That's the version of this story I want to help tell — not artificial intelligence against us, but the other AI, working alongside the one we've had all along.