Seniority Buys Time, Not Immunity

August 10, 2026   By now you know I believe AI is going to have a sweeping impact on the labor market. Long term, net positive. Short term, disruptive.

A while back I wrote about the Stanford study showing AI's employment effects were hitting 22-25 year olds in AI-exposed jobs while everyone else kept chugging along. I wondered if these were the early impacted (instead of early adopters) and might be the canaries in the coal mine.

Bad news. A Stanford researcher named Mary Kate Stimmler just re-ran that analysis on the public data through June 2026, and the canaries have company.

In the most AI-exposed occupations, four of the six age bands are now past their employment peaks. And here's the part that made me put my coffee down: they turned in order of age.

• 22-25 peaked first, in late 2022. Now sitting at 88 on the index.

• 26-30 peaked November 2024. Now below where they were when ChatGPT launched.

• 31-34 peaked the same month. Down 3+ points since.

• Even the 50+ crowd has been sliding since late 2025.
  
  

The chart isn't showing six different stories. It's showing one generalist climbing the corporate ladder.

AI is a generalist. It doesn't attack your job title, it slides underneath your workflow as a base layer and does the general-purpose stuff first...the research, the first drafts, the summaries, the routine analysis. Then it climbs. Every capability jump takes it one more rung up the experience ladder, absorbing work that used to require another year or two of reps.

For some sectors, like Information, the layoffs are accelerating. The Kobeissi Letter points out the layoff rate in the information sector jumped +0.7 percentage points in June, to 2.3%, the 3rd-highest since the 2020 pandemic.

• Since November 2025, the layoff rate has more than doubled.
• In absolute terms, 63,000 workers were laid off or discharged in June, the 3rd-highest monthly total since April 2020.

As a result, the 6-month moving average layoff rate rose to 2.0%, the 2nd-highest on record.

By comparison, this metric peaked at ~1.5% during the 2008 Financial Crisis and the 2001 recession.

Now, before anyone panics, let me put my skeptic hat on, because the aggregate data tells a calmer story. The same Stanford institute published a brief showing unemployment for the most AI-exposed workers has actually risen slightly less than for the least exposed since 2022. And a Ramp study of 21,000 companies found the heaviest AI adopters grew headcount 10% over two years, and grew entry-level hiring 12%.

So which is it? Apocalypse or nothingburger?

Neither. And the reconciliation is something I wrote about a few weeks ago.

Very few are getting fired by AI. They're just quietly not getting hired.

Firms are frozen in the same question I've been wrestling with all year: what will AI actually do to this role? Until they know, the safest move is to not backfill. The senior people keep their jobs. The next opening just never gets posted.

Cisco is rolling out a personal AI agent to all 90,000 employees this quarter, and their CFO says AI already writes 80-90% of the first draft of the MD&A section in their public filings. Sounds an awful lot like an Offering Memorandum, doesn’t it? Ask yourself: who used to write that first draft?

A 24 year old. That's who.

The reason seniority buys time is the same reason it was ever worth paying for. The junior job is mostly repeatable tasks with a little judgement sprinkled in. The senior job is mostly judgement with repeatable tasks sprinkled in. The generalist eats from the bottom up, in age order, exactly like the chart shows. But time is all it buys.

The chart I showed you earlier doesn't show a line where the climbing stops.

It shows a schedule.

I wrote about our actual roles that AI eliminated, but I think the far more common outcome will be reallocating instead of replacing. Mercer found it cost Standard Chartered $49,000 less to move an existing employee into an emerging role than to hire externally. Technical skills can be taught when AI makes it easy to code. What you can't hire off the street is someone who already knows your systems, your clients, your unwritten rules.

Which brings me to the advice I give my team and my kids, and it's the same three words for both:

Make yourself indispensable.

Not to a task list...the generalist is coming for the task list, one rung at a time. Indispensable to the judgement. Be the person AI hands its exceptions to, and the schedule in that chart stops applying to you.

And not one part of me believes that chart will stop just short of wherever I reside. I know it is, which is why we went all in on AI in March.