Two people use the same AI. One now gets through dozens, even hundreds of times more work than before. The other barely changes, and some people are even seeing their work disappear altogether. Since AI spread, that gap has become plain to see in so-called white-collar workplaces, starting with IT engineering.
In jobs where you move your body on site, or work face to face with people, the relationship with AI will likely be different. What I am writing about here is white-collar work.
AI is supposed to be a tool anyone can use the same way. So why does such a gap open up? For a long time I could not put it into words. Recently I have come to think the key word is "speed."
Fast people can see the goal
The people whose work has taken off with AI were fast workers to begin with.
Why are they fast? Probably because they can see the goal before they start running. They know where to head, so they can start right away, and they get there in a straight line without detours.
Slow people cannot see the goal when they start. So they get lost along the way. While lost, they wander down side paths, come back, and get lost again. It is not that each step is slow; not seeing the goal is what ends up making them slow.
Until now, the fast had to wait
Until now, though, a difference in speed might produce a gap of a few times, but never a gap of dozens or hundreds of times.
However fast you ran, work was done in organizations. As long as it was teamwork, fast people had to wait for slow people somewhere along the line. Sitting through unproductive meetings, waiting for sign-off, waiting for replies. You could say the organization itself kept the speed gap small.
On top of that, a person has only one body, and needs time to sleep. However fast someone was, there was a limit to how many hours they could work in a day.
AI doesn't make you wait
AI never makes you wait.
Fast people who can see the goal give AI one instruction after another, check what comes back, and move straight on. All the time they used to spend waiting for someone else simply disappears. Fast people are accelerating even further with AI. That is where we are now.
What is more, AI never sleeps. Give it instructions at night before bed, and the work moves forward while you sleep. Run several AI agents at the same time, and one person can push forward the work of several people in parallel. With AI, even the physical wall of having only one body can be overcome.
Slow people, on the other hand, do not get there even with AI. AI gives you what you ask for. But deciding where to go is up to the person using it. If you cannot see the goal yourself, you do not even know what to ask AI for. However fast AI runs, if there is no destination, it arrives nowhere.
The era when organizations absorbed the speed gap is coming to an end. We have entered an era in which fast people win everything.
Research that says AI helps the less skilled
That said, there is also research pointing the other way: that AI actually helps less skilled workers the most. A study of about 5,000 customer support agents in the United States found that with AI assistance, issues resolved per hour rose 14% on average and 34% for less experienced agents, while experienced agents saw almost no effect.
But that applies when the level of thinking required is shallow. When the shape of the answer is more or less set and you can move forward without thinking deeply, AI can show the path to the goal, so slow people can catch up. When deep thinking is needed, such as picturing the goal itself, that does not happen.
The ability to picture the goal
Then is "being able to see the goal," which separates fast and slow people, something you are born with? It is not.
Being able to see the goal takes long training and experience in continually using your head. You think for yourself and reach the goal. Only after doing that many times can you see the goal before you start running.
Can't children start building that experience while they are young? That is what I hope for. This kind of experience does not have to come from a computer. But for now, I believe a computer is the most efficient way to build it. Using a computer, deciding for yourself what to make, working out how to make good use of AI along the way, and making it toward a goal. I believe that experience lays the foundation for the ability to picture the goal.
*
At Digital Kodomo BASE's Houkago Lab, children decide what they want to make and turn it into games and other works, using AI along the way. They put what they want into words, tell the AI, say it again when it comes out wrong, and see it through to the end. Through that repetition, we hope children will become able to picture their own goals.