What follows is one unbroken chain of association, recorded in one sitting, on a subject nobody has ever paid me to have an opinion about. Two lines have been cut. The dead ends have not been, because the dead ends are the evidence.
Everything set against the red rule is verbatim. The small grey text is scaffolding and it argues nothing. The large centred lines are my compression of the jumps, not his words.
An AI diagnosing Parkinson's from a scan before a person can see anything there at all. Pattern recognition outside human sight.
Then, without a bridge:
The way we communicate, our language, is the reason we cannot be more intelligent than we are.
That is the premise of Arrival, used as a constraint model for machine intelligence rather than as a film reference.
If a mind were not bounded by our language, it might simply hand us climate change.
This is fascinating because I feel like, as a human, I am quite good at that.
He has just named his own ability and the thing that would make it obsolete, in the same breath, and then carried on.
I think the answer is in that question.
He does not answer it. He builds the question so the other person cannot avoid answering it, which is the whole method in one line.
Someone saying, did you know you can give this single prompt and it will build an entire marketing campaign.
You and I know this is not true, because you need the skills. We have had some frustrations. We have had some run-ins. You have forgotten things. I have said things.
It is a relationship you need to build.
The "you" there is the tool. Anyone claiming AI fluency on a website in 2026 is claiming what everyone claims. Naming the friction is the only version that is not slop.
Right now the conversation is whether AI is going to steal your job. That is not where the energy should go.
Filed under threat to employment. He renames it as a question about what a species does with freed time, and whether the economic system distributes that freedom or concentrates it.
Same operation as a churn number filed under support. Different subject, considerably larger stakes.
Two futures, held at once and not resolved. One where the work is done and the freedom is distributed. One where it is concentrated in a small number of people in a small number of places and everything else is commoditised, which ends in barter.
He does not choose between them. He asks why nobody is having the conversation.
And if the systems were fair, would an AI tell a toddler what they should be?
That is where it stops. No resolution, no lesson, no call to action. Six jumps, three dead ends, and it lands on a question.
Ideas almost always come over food and in the shower. I journal coming out of the shower, or voice-to-text on a dog walk. Those are the moments where thinking is not polluted.
Work has a lot of things that ruin that context. Slack, notifications, everything competing for attention.
The best ideas come from incredible stillness, or a large amount of play.
And the part that is a diagnosis rather than a preference:
If you have a globally distributed team in different time zones who never see each other, it is really hard to be creative in your own echo chamber.
You might generate the idea. That is the shower. But having it evolved, pollinated and built on thrives in social interaction.
He has a specific company in mind and I am not going to name it. The point is that this is falsifiable: it is an explanation for why one organization could not innovate, not a view about offsites.
There are dreamers, designers and architects who need to meet regularly to make step changes. And there are people who are builders, and are happy being builders.
A lot of people work to live. They do not live to work. Organisational design has to account for that.
Most people who believe in creative collaboration assume everyone wants in. He designs for both and says so out loud, which is the version with an edge on it.