The Waymo Effect: How AI Is Quietly Making Research Less Collaborative

I work at an intersection of tech, applied research, and science.

Something I’ve noticed in collaboration that does occur is an increased confidence in people outside their domains to say things with conviction. I have people who have limited experience with software pushing out layers and layers of abstracted code that’s fairly sophisticated but often misguided in intent who will say what they’re doing is correct, with conviction.

I also hear a lot more questioning people in their domains and challenging opinions, then hearing what I can only imagine are fragmented pieces of conversations they had with an LLM thinking through some argument. Then there’s silence when you discuss shortcomings, then they come back later with their memorized fragments of what you said, combined with memorized fragments of the LLM response to the argument.

It’s occurring, a lot more. People are treating their LLMs in collaboration as a source of truth and using then to focus on their specific path or goals they think or have bias towards going down, vs just opening discussing things, considering tradeoffs from experts multiple disciplines weigh in on and then taking an approach that everyone finds most agreeable.

It’s making me want to be a lot less collaborative with such individuals. I don’t want to sit around and refute Claude text outputs all day.

AI is extremely useful, but it’s also extremely easy to fool yourself into thinking you understand what is going on without really understanding. This is often true with the code, but also for math and science concepts, etc.

Not many professions are formally trained to be cognizant of this lack of understanding, and how to confront it.

Usage of AI in collaborative settings is an amplifier of these issues, especially if someone doesn’t realize they don’t understand: they couldn’t teach or explain the concepts they use, or be forced to work with them malleably in a way that an expert or researcher would.

If you are cognizant of your lack of understanding, you can remedy it by slowing down and teaching yourself. This is required to make better use of AI in the domain of interest!

But you can’t have all things at once: you can’t move at speed with AI, collaborate effectively, and understand what is going on as an expert would. It is not physically possible for a human brain.

I wonder if AI is making us dumber

I can have AI pull some analysis - sql query, crunch some numbers and bam, some precise sounding number.

But then I question it, and then it says "you are right, I didn't check that"

How did I know what to question? By doing the friction work of querying myself, calculating myself, and then getting questions from leadership.

I worry about the younger generation. They spit out Claude results fast, people question it and they say "ok I'll take that back to Claude".

I suppose they'll eventually learn to do their own sense checking because they still have the friction of presenting to others.

And maybe my model is a thing of the past. Everyone is now going to be a manager (of agents) fresh out of school. They aren't doing the work anymore but they will need to set clear goals, validate work and hold agents accountable.

The notion that "every human interaction is worth preserving" is probably not right, and Waymos are showcasing this. As the author mentions, many people don't enjoy the forced small talk with Uber drivers. Interacting with people from different backgrounds and perspectives is important, but I doubt the forced interactions like taxi rides are where we were getting those until day.

I think the problem is the same as with any technology: used correctly it adds

value, used incorrectly it subtracts from it.

LLMs can massively speed up a process, but without control they can turn into

social "sources of truth." A research process has well-defined, well-founded

phases: you delimit the topic, search for sources, evaluate whether they're

suitable, review their content, and place them on the map of the subject. The

more sources, the greater the knowledge, and the better the final result —

mental, or in the form of a report — is built. For that you have to read, and

reread, and think, connect, relate, and conclude.

AI can do all of those steps faster than a human. But if we let it do the

entire job on its own, its own way, with no checks at each stage, the

conclusions can end up distorted. If we know what we want it to do and how we

want it done, and we put the mechanisms in place to enforce that, the result

is different — better, more reliable. It's worth remembering: it's just a

tool, nothing more.

The wording of this comment in English has been corrected with AI, I don't

have enough fluency to express myself clearly, but I do review the final

result. In this case it got the verb tenses wrong, I saw it clearly, but the

AI didn't understand it, it took me several instructions to explain it so it

would understand. It's a tool, without supervision it can lead to problems,

but it has expanded the world for a lot of people.

I'm seeing this on a current academic research project as well- it's like throwing fuel on the fire for all the best and worst parts of working with researchers.

First: it's definitely become harder to get people to integrate their work and use standard tools. We're exploring creating AI skills etc for our tooling, simply because the actual audience we need to convince isn't in the room. If the LLM doesn't echo our recommendations, people will go with whatever one off script their personal bad idea bear churns out. Many research software errors are edge cases (scaling, numerical errors, etc), and "works on my computer" syndrome is endemic in the literature. The field has made big strides in enabling computational reproducibility, but lately it feels like we're set to lose ground again.

Second: LLMs have a bias to action, and can easily bury the user reporting on whatever. I've seen multiple seminars recently where the Q&A devolves to "Q: What are the implications of this finding? A: I don't know, this is just presenting a report on the results".

It seems that humans are still figuring out how to maintain agency and steer the chatbot to the big picture, and unreviewable science is just as likely an outcome as unreviewable code... but with far less automated tooling to help guide the process.

Historically, PIs nominally guided the big picture, and the entity who did the work was a participant in the review process; now students are having to look at projects from a new angle wi […]

I think this notion of friction vs frictionless is one of the key features of this time in history.

To my eye, the whole line from personal technology in the 80s -> through UX on the internet -> through app platform lock-in -> through subscription-centric "ownership" have been providing us all with the promise that we can buy our way into less friction. The next device or subscription solves problems or enables us one way or another. We see this line of history, draw it into the future and see a utopian future with zero friction.

Zero friction meaning zero difference between how you want to feel about the world and what you get out of it. More feeling empowered or satisfied or comfortable or right.

The parts of our culture that embrace friction are the slow parts - grousing with neighbors, weeding a garden, letting the other car go first. These are the parts where we meet with the reality of other people or nature. Maybe there's a tool or framework or gadget that could help, but we know and accept that there is something immutable outside of ourselves and we have to find a way to flow with it.

I'm getting older, I'm finally starting to see the limits of pursuing a frictionless life.