The slow clock decides it
Every technology runs on two clocks. One is the clock of announcement, which moves quickly and is easy to watch. The other is the clock of ordinary adoption, which moves slowly, is hard to watch and decides almost everything. Most misunderstandings about artificial intelligence live in the gap between them.
The slow clock is not a failure of the fast one. It measures something different: the time it takes for a capability to become a habit, for a habit to become a standard, and for a standard to become something nobody has to think about. Each step takes a season or two, and none of them can be shortened by enthusiasm.
The gap is also where most of the disappointment lives. A capability described as finished on the day it appeared is judged a failure eighteen months later, when it is simply somewhere in the middle of the slow clock, doing what everything else has done before it.
Five things worth waiting for
Waiting is easier when it has a subject. These five changes tend to arrive late and matter most, and each is worth more attention than the announcement that came before it.
None of them can be scheduled, and all of them can be recognised at the time by anyone who is looking.
- Tools that have survived a full year of ordinary use, including the weeks when nobody was excited about them.
- Explanations plain enough that someone outside the field can repeat them accurately.
- Results that hold up when the person checking them is not the person who produced them.
- Habits that outlast the enthusiasm of whoever introduced them.
- Quiet agreement between people in different places who arrived at the same practice separately.
Patience is not passivity
None of this is an argument for waiting to see. Patience of the useful kind is busy. It tries things early and small, keeps notes, repeats what worked, drops what did not, and refuses to draw a conclusion from a single good week.
It also tolerates being unimpressed for long stretches, which is harder than it sounds in a field that produces something impressive most months. The reward is a view of what is happening that holds steady across a year rather than a quarter.
Practised across a season rather than a week, patience turns into a kind of evidence. The notes accumulate, the repeated results begin to agree with each other, and a view forms that was not borrowed from anybody.
October, with three months still in front of it, is a reasonable place to practise that.
Frequently asked questions
What are the two clocks a technology runs on?
The clock of announcement, which moves quickly and is easy to watch, and the clock of ordinary adoption, which moves slowly and decides almost everything.
What is worth waiting for?
Tools that survive a year of ordinary use, explanations plain enough to be repeated accurately, results that hold up under someone else's checking, and habits that outlast the person who introduced them.
Is patience the same as waiting to see?
No. Patience of the useful kind is busy: it tries things early and small, keeps notes, repeats what worked and refuses to draw a conclusion from a single good week.