042: What AI Does to Your Salary

Hi there

Your pay might not go up for the rest of the decade. And the economy will be doing great.

That is the line I could not get past last week, reading Anthropic's new model of the economy up to 2030. Here is the link.

(Anthropic makes Claude. Yes, the AI company modelled what AI does to your salary.)

Time to read: 7 minutes

In this edition:

1. Twice the Growth, No Raise

2. Who Gets the Money, and Who Moves

3. Who Keeps the Job

Speakers Wanted: Claude for Leaders

We are looking for speakers for Claude for Leaders in Zürich. Non-technical speakers, non-technical audience. This time one condition: your company is based in Zürich or the area, because the evening is in person. If you use Claude in your own work and have a real story from inside your company, send me a message.

Twice the Growth, No Raise

Imagine it is 2030. The economy is growing twice as fast as it does today. Faster than in the best year of the dot-com boom.

And your salary is not growing.

That is the middle of the three scenarios in the model, the one most people in Anthropic's survey expect.

In it, AI can do about half of all knowledge work. Companies use it for a good part of that. Double growth is happening.

For knowledge workers, pay keeps growing at the usual slow pace. AI adds nothing to it.

Everyone else gets the usual raises, plus around 6 percent on top.

Knowledge worker, in the report, means management, professional, sales and office jobs. Nearly two out of three people with a job. Most of the people reading this. Understand: if you have a job that you do from behind a computer, you don't benefit from the huge economic boom.

And no, nobody gets fired in this picture. Well, almost nobody. Unemployment for knowledge workers goes up by about 1.5 points. You keep your job. Just the value of your job doesn't grow.

At this point, you may ask: so where does the money go?

To the owners of the machines. Four cents of every dollar move from pay to capital in four years. That is nearly the whole shift the US saw in the forty years after 1980. (I read that twice.)

One more thing. None of this shows yet. The model says almost all of it starts after 2027. Right now, it's the calm before the storm.

(For the record, the scary scenario: knowledge workers earn 11.5 percent less, and almost one in five of them is out of work.)

In case you are wondering how Anthropic came up with this: they asked almost 11,000 Americans what they expect AI to be able to do by 2030, how much people will use it, and how fast someone finds a new job after losing one to AI. Their answers went into a model built by economists and checked by 18 outside economists. Remember, they know exactly who uses AI for what. The authors say it themselves: these are not predictions. Fair enough. I still cannot get past that salary line.

Who Gets the Money, and Who Moves

If knowledge workers do not get the raise, who does?

Electricians. Construction workers. Caretakers. Entertainers. Anyone whose work AI cannot do.

The report's own example: when AI draws up building plans and gets the permits through faster, more buildings get built. More buildings, more construction workers, higher pay for them.

So some people change sides. Knowledge jobs shrink by about 4 percent, everything else grows by about 5 percent.

And that move is hard. The economists say it in one line: a software engineer does not easily become an electrician. Finding a new job takes long already now, even if you just want to continue what you did before. In Switzerland, for senior roles often 9-12 months. Now add to that the time for transitioning your profession...

Another part I find especially interesting.

Of the tasks AI touches, three quarters get taken over by AI completely. One quarter stays with a person, working with AI (human-in-the-loop). And for every four tasks handed over, one new task appears that only a person can do.

Pay follows the tasks that stay with people. Checking. Deciding. The new stuff. The ones that need you to be a human.

If everything you do today can be handed over to AI, your job goes with it. Then you are the one looking for new work.

(You have heard me say it on stage: give away the work, never the judgment. Now there is a wage model behind it.)

Oh, and my favourite number in the whole thing.

The knowledge workers in the survey were asked how much of their own working time AI could do by 2030. Typical answer: 27 percent.

The same people, shown a list of eight normal office tasks (emails, bookkeeping, a tax return, a contract, a piece of software), said AI could do 71 percent of them.

Always somebody else's job.

Who Keeps the Job

I cannot tell you which scenario arrives. Nobody can.

What I can tell you is who stays on the knowledge side, in every one of them.

The model runs on averages. It has no idea who ends up with the flat pay and who does not. But the logic inside it is clear. The people who stay are the ones who hand the routine work to AI and keep the checking, the deciding and the new tasks for themselves.

In other words, the ones who are good enough with AI to know what to hand over and what to keep.

That is a skill. And very few of us have ever measured it.

It is about more than getting faster, by the way. Speed is the boring part. The people who are good with AI do things they could not do before. They build a tool for their team in an afternoon. They test a business idea over a weekend instead of writing a proposal about it. They find a cost nobody had questioned, or a service nobody had thought to offer. New work, new revenue, new ways of running the business. That is where the pay follows.

So, one question. How good are you with these tools, honestly? And can you change your mindset and work differently?

I built a short check for exactly that. Ten statements about your own use of AI, two minutes, and honestly a bit of fun.

Your result lands in your inbox, with a plain reading of where you stand and what to do next.

If the results show a gap you want to close, that is the work I do one on one with a select group of leaders or business owners, that are turning their ways of working upside down (for the better).

Someone I started working with recently just got a chief of staff, and began changing how her business runs after the first two hours we spent together.

I only have very limited availability for this. Book a short call if you want to apply.

Talk soon,

P.S. The model is public and you can play with it. Put in your own assumptions and watch the salary line move. The link is at the top.

How can we work together?

• You have rolled out Copilot. Your teams have gone through hundreds of training sessions. But in their daily work? Nothing's really changed. Take my free, 5-min assessment to see your AI leadership readiness.

• Hire me as your executive coach to help you master change. Sign up for a 30-min complimentary coaching call (waitlist applies).
• Book me as a speaker, panelist or moderator for your event. Get some impressions on how it is, here. Or watch my TEDx talk here.

Did you miss our last newsletters?

Inboxes can get a little crazy, right? Just in case you missed them, here are the links to our previous newsletters: