The Waymo effect: an essay warns AI is quietly making research less collaborative

Riding a driverless car in San Francisco gave Daniel Hook, chief scientific officer of the Holtzbrinck group, an uncomfortable thought about artificial intelligence and research. In an essay for Research Agenda, he recalls climbing into the back seat of a Waymo and, with a colleague who had crossed an ocean to see him, admitting the ride was wonderful.

Naming the Waymo effect

Hook was in San Francisco partly to work with Susan Winslow, CEO of Macmillan Learning; both work for Holtzbrinck, both are introverts. The driverless car was a small oasis: no obligation to make conversation, a guilt-free space to finish an email. The discomfort came later: two people who had crossed continents to talk to each other were delighted by a technology whose main feature is that you need not talk to anyone.

Hook proposes a name for this. The Waymo effect, he writes, is what happens when a technology removes the friction of dealing with another human being and we experience that removal as pure gain — because the costs of friction were always visible, while its benefits were not. He connects it to Tim Wu’s “tyranny of convenience” and Albert Borgmann’s “device paradigm”.

The frictionless colleague

Large language models, Hook argues, are the Waymo of intellectual life: a collaborator appears occasionally, between teaching and deadlines, while an LLM is available at 2am on a Sunday. A collaborator brings their own agenda and the inconvenient conviction that your central assumption is wrong; an LLM challenges you exactly as robustly as you request, and never tells you, unbidden, that you are solving the wrong problem. “The collaborator’s inconvenience is not a bug in the collaboration; it largely is the collaboration,” he writes. For the erosion that follows, he proposes a word: decollaboration.

The incentive trap

The essay is not about temptation. Funding pressure kills the unmeasurable first — travel, workshops, sabbaticals, “the entire physical infrastructure of serendipity” — while evaluation systems still reward speed and output above all. And an LLM never argues about author order and claims no share of the credit. Where credit is the currency of survival, that is not merely convenient; it is arbitrage.

Writing is thinking, so fund the friction

Hook also attacks the promise of speed: writing takes time because thinking takes time, and outsourcing the writing does not accelerate thinking, it skips it. He cites psychologist Robert Bjork’s “desirable difficulties” — struggle, not comfort, is what builds durable understanding — and warns that a research culture can look healthy by every visible measure while what sustained it falls below threshold.

Hook cites a March 2026 Nature comment by Dashun Wang, who called AI agents “aeroplanes for the mind” and asked for “pilot-in-command science”. His prescription: fund the friction — the workshops, visits and unstructured time that produce conversations nobody could have scheduled — and treat “who did you think with?” as seriously as “what did you publish?”