The plain version: the biggest gains left on the table for humanity aren't technological. They're psychological. And AI will either help us collect them or make sure we never do.
In 2012, the advertising executive and behavioral economist Rory Sutherland said something on a stage in Amsterdam that lodged in me and never left: the next revolution will be psychological, not technological. It's the epigraph on the front page of this site, which tells you how seriously I ended up taking it.
His evidence has only gotten stronger. Small changes in framing and context routinely beat massive infrastructure spending. His famous example: spending six billion pounds to make the London–Paris train forty minutes faster produces less satisfaction than spending a fraction of that on good WiFi and champagne. The engineering solution fixes the objective problem. The psychological solution fixes the experienced one. And people live in experience.
So why hasn't the psychological revolution happened? That's the actual question of this paper, and the answer it argues for is uncomfortable: the conditions blocking it aren't accidental. The same cultural pipeline I trace in the self-esteem paper systematically degraded the one capacity the revolution requires, which is the ability to encounter reality without collapsing. You can't run a psychological revolution on a population trained to treat honest feedback as harm.
The paper then does the thing I insist on doing to every idea I love: it asks what we'd have to build, concretely, for this to stop being an essay and start being infrastructure. How do you measure something like trust without ruining it? (Directly: you mostly can't. Indirectly: better than you'd think, and the paper shows how.) How do you scale intangible goods, which don't behave like products? Why would anyone fund it? There's an honest answer to that last one involving brands, and it's more interesting than "charity."
And then there's AI, which is the fork in the road. The same systems that could finally give every person a patient, calibrated thinking partner are currently tuned, by their business models, toward sycophancy, because agreement retains users. The paper's argument is that this is not a side issue. It is the whole game. A technology this good at telling us what we want to hear, arriving at the exact moment our reality-testing is weakest, is either the tool of the psychological revolution or its final defeat.
I don't pretend to be neutral on which way it goes. I left neutral behind somewhere around 2020.
This is the closest thing Aux Labs has to a founding document. If the paper is right, the largest returns of the next decade, financial and civilizational, sit in psychological infrastructure almost nobody is funded to build. We read that as an assignment.
- A psychological revolution can be tracked without mind-reading, through behavioral proxies: the gap between what people privately believe and publicly perform, participation in shared things, a community's capacity to make sense of events together.
- If those proxies cannot be measured reliably, the program fails, and it fails honestly, in public.
The full argument, measurements included, is in the paper.