# The Waymo effect: who is driving your research? > A driverless ride felt like pure gain until two colleagues noticed what it quietly removed, and realised AI is doing the same thing to research. - Title: The Waymo effect: who is driving your research? | Shortify - Summary: A driverless ride felt like pure gain until two colleagues noticed what it quietly removed, and realised AI is doing the same thing to research. Friction… - Keywords: research culture, collaboration, automation, incentives, technology, startups, Waymo, effect, driving, research, Research Agenda - Source: Research Agenda — https://researchagenda.news/articles/the-waymo-effect.html - Author: Daniel Hook - Published: 2026-09-07T00:00:00+00:00 - Read time: 4 min - Topics: ai, research culture, collaboration, automation, incentives, technology, startups ## A car with nobody in it Daniel Hook summons a Waymo in San Francisco, climbs in, and nods politely at an empty driver's seat. The steering wheel turns itself the whole way. > I summoned a car with no one in it. ## Two introverts, one oasis He and Macmillan Learning CEO Susan Winslow had crossed an ocean to collaborate in person. Comparing notes, they agreed the best part of the ride was not having to talk to anyone. ## Naming the Waymo effect A technology removes the friction of dealing with another human, and we feel it as pure gain. The costs of that friction were always obvious. Its benefits never announced themselves. > the costs of the friction were always visible to us, while its benefits were not ## Nobody advertised the friction Tim Wu called it the tyranny of convenience. Albert Borgmann went further: a device hands you the commodity while hiding the practice that used to earn it, so you stop noticing the practice is gone. > once a frictionless option exists, we take it by default ## The last unchosen conversation The taxi driver was often the final person outside your bubble you were obliged to speak to. The last reliable source of an opinion you did not go looking for. > the last stranger we were obliged to encounter ## Lost one ride at a time Neither of them would design a world without those conversations. They just enjoyed skipping this one. That is how such things disappear. > never by decision, always by convenience. One comfortable ride at a time. ## The frictionless colleague A collaborator is available between teaching, grants and time zones. An LLM is available at 2am on a Sunday, with no agenda beyond being useful to you. > A collaborator must be persuaded; an LLM must merely be prompted. ## It critiques what you brought Ask a model to attack your argument and it will, capably. It won't volunteer that you're solving the wrong problem, or that a rival group tried this in 2019 and gave up. > Large language models are the Waymo of intellectual life. ## The inconvenience is the point A co-author's refusal to simply agree with you isn't a defect in the collaboration. It is most of what the collaboration actually is. > The value of another mind lies exactly in the ways it refuses to be an extension of your own. ## A word for it: decollaboration Research isn't a machine for turning ideas into papers. It's a community held together by arguments, apprenticeships and conference-bar talk. Every chat redirected to a chatbot pulls one thread out. > No individual thread matters. The fabric does. ## The incentive trap Collaboration has always cost travel, scheduling, ego management and the small heartbreak of author-order talks. Publish-or-perish taxes every hour of it. Three forces are now compounding that tax. ## We defund the corridor When budgets tighten, the first cuts land on travel, workshops, sabbaticals and visiting positions: everything whose value is real but unmeasurable. > We defund the corridor and then wonder where the corridor conversations went. ## Velocity worship Whatever DORA-style statements sit on our websites, evaluation still rewards output and speed. The models promise a literature review by Friday and a draft by Monday. ## Credit arbitrage The model never argues about author order. In a system where credit is the currency of survival, a sharp interlocutor who wants none of it isn't just convenient. > It is arbitrage. ## Small teams disrupt Team-science research shows small teams disrupt while large ones develop, and that odd pairings across fields produce the most novel work. Early AI evidence points the other way. > individual productivity rises while the diversity of ideas narrows ## Writing is thinking The paper was never the point. Writing is where you find the hole in section three, where a vague intuition either sharpens into a claim or dissolves. > Outsource the writing and you have not accelerated the thinking; you have skipped it. ## Desirable difficulties Robert Bjork found that effortful retrieval and the struggle to get an idea into your own words aren't obstacles to understanding. They're how understanding forms. > Remove the difficulty for the sake of comfort, and you do not get the same understanding faster. ## Healthy until it breaks Hook compares it to PT-symmetric quantum systems, which behave normally right up to the moment a hidden symmetry snaps. Outputs up, turnaround down, prose immaculate, foundation gone. > By the time the symptom appears in the metrics, the cause is years in the past. ## Pilot in command Dashun Wang calls AI agents aeroplanes for the mind and insists the researcher stays captain, owning the question and the conclusions. His fix for converging outputs is to design for dissent. > the researcher as captain, the agents as crew ## Who is in the front seat? Lisanne Bainbridge warned in 1983 that skill erodes exactly where automation is most reliable. So fund the friction, judge contribution over velocity, and ask who someone thought with. > And when did we stop noticing that nobody was driving? ## Key takeaway Friction with other people isn't a cost of research, it's the mechanism. Fund it on purpose, or convenience will delete it one comfortable ride at a time.