Howdy, picnickers!
Last week we talked about dashboards no one uses and reports no one reads. This is the perfect segue into an essay I’ve been meaning to cover for a very long time:
As we stumbled into the digital age, the promise of interactivity turned sour. It turned out that while it was nice that you could interact with stuff, having to do so became a slog. So products appeared that could free you from having to use them. A taped show or a photocopied article would sit on the shelf forever, freeing you from the obligation to watch it or to read it (a more modern example might be your inbox chock-full of unread newsletters).
LLMs represent a new frontier of interpassivity. Now we can make a robot read the article for us; the pleasure is not merely deferred indefinitely but farmed out entirely.
The academics who photocopy an article to read later are indulging in an unconscious fantasy of reading, but they would never tell you they had actually read it. But Generative AI platforms center the idea that one has written, or created, by interacting with a structure they did not build and do not understand.
This shows up most acutely in vibe coding. The systems (both the “agentic harnesses” themselves and the hype whipped up around them) take great pains to paint the user as the programmer, the active participant who is merely engaged in distributed cognition on an unprecedented scale. But in reality, Claude Code is a consumer product, and vibe coding is an act of consumption.
And that relationship is trending further towards passive consumption. Now, “reasoning” models even talk to themselves for you: let Claude Code rip and it will happily describe all the things it’s doing on your behalf; your opinion is superfluous. It is nothing like pair programming — the robot calls the shots and you disengage your brain:
“Agentic” might sound more advanced, but it actually doesn’t mean anything besides making you wait longer and giving you fewer opportunities to provide input.
AI did not usher in the phenomenon of interpassivity. It emerged under the flickering glow of the very first blinkenlights. But it has scaled the problem up to an unmanageable level, like it has every other problem in the industry.
The crisis that UX finds itself in, for instance, is also nothing new. We were already overwhelmed by cross-functional partners agitating for the feature factory over system thinking. The ability to trivially generate a functional demo faster than you could think about what you were doing just pushed teams further over the edge:
I’m not convinced the “AI” is doing much of the work. The permission to go full-feature-factory is where the gains are coming from. When the tools enable “I have a cool idea” to become “my idea is validated by research” without understanding the situational context of the user, things will dramatically speed up. The team will appear more productive. But what is getting better? For whom?
The feature factory approach was always pure self-indulgence: living out the fantasy that we could simply know the right thing to build, knock it out of the park on the first try, and then move on to the next feature — and somehow they would all add up to inevitable success. That we could get there from here, and the only thing stopping us was velocity.
That was the logic behind UX theater that diluted real research practice a decade ago. It has now returned with a vengeance to AI-ifiy what little rigor remains:
The AI-ification of research seems to me to be amplifying many of the shifts that I noted back in 2017.
But the pleasure interpassivity can bring you is never going to be as good as the real thing. The industry has managed to displace the costs of playing make-believe. Some of those costs are displaced into the future, in the form of enormous maintenance costs involved in keeping systems coherent two, five, ten years from now. Other costs are displaced onto employees — such as the permanent hypervigilance demanded from any operator of tools that fail in unpredictable ways.
That is the ideal environment for interpassivity to take root. Too much is asked of us. We crave being able to disengage, to say “no.”
Participation is the dominant mode and expectation of culture, and it‘s exhausting. It's everywhere and overwhelming: too much to look at, too many places to be, too many demands to produce and engage…Interpassivity is wrapped up in a slim fantasy, a daydream that gives us just enough permission to say no to doing more.
It feels like we have two choices: to give up on the pleasure of doing a good job and replace it with the ersatz alternative, or we burn out trying to hold the line against low-quality garbage being shoved out the door.
The only cure is to find those other people who still care, and still choose activity over passivity. Be where the crafters are. Refuse to let your brain atrophy.
You'll end up ahead of the pack, as the rest of the field painstakingly rediscovers that rigor and process were never optional, and cannot be substituted with a software tool.
— Pavel at the Product Picnic