Why companies are wrong about AI layoffs

The narrative is everywhere. AI can write code. AI can draft emails. AI can design interfaces. Therefore - the logic goes - you need fewer people. Cut headcount, keep the tool, save money.

I hold myself to a high standard, and I still find people surprising me - steering AI to deliver work at a level I did not think was possible. That is the signal companies are missing.

The layoff logic rests on a fatal assumption: that AI output is good enough. It is, if your standard is average. But average does not win markets, does not build products people love and does not compound into anything defensible. The people who know how to push AI past average are the most valuable employees you have ever had. And companies are firing them.

01AI is the new baseline, not the new ceiling

AI learned from the whole internet - brilliant papers and terrible blog posts alike. It internalised the full bell curve of human output and, by default, hands you the peak: the most statistically likely response. The average.

That is genuinely useful. The average is competent. It handles boilerplate, first drafts and common patterns, and anyone can now produce decent work in any field with no training. That is a real shift.

But decent just became table stakes. The differentiator moved up, and the gap between AI-generated average and expert-guided excellence is enormous - and growing.

02Expertise became more valuable, not less

Here is what the layoff narrative gets backwards. AI does not diminish the value of expertise, it amplifies it. The tool is identical for everyone. The difference is who is holding it.

You have to know what excellent looks like to steer AI past average. That is not a skill AI has. It is a skill people spend years developing.

03The layoff maths runs backwards

When a company lays off 30% of its workforce "because AI", it is making a bet: that AI-generated average output is enough to compete, and that cost reduction matters more than quality differentiation.

Here is the problem with that bet. The competitors who kept their best people now have a workforce where every person operates at three to five times their previous output, at the same quality bar or higher. You saved salaries. They multiplied capability.

One company optimized for the cost of average output. The other optimized for the ceiling of excellent output. This is not a close race.

04The right tail belongs to people

I keep getting surprised - not by AI output, which is predictable by now, but by what people do with it. People who deeply understand their craft use AI as a force multiplier for their judgement, not a replacement for it.

They know what to ask. They know when the output is subtly wrong. They know how to iterate in directions the model would never choose on its own. They are not using AI to do their thinking; they are using it to execute their thinking at a speed and scope that was not previously possible.

These people do not just use AI. They bend the bell curve; they shift the whole distribution toward excellent. And they are the ones being let go because a spreadsheet says AI can do their job.

05The winning move is the opposite

The companies that dominate the next decade will not be the ones cutting people. They will be the ones investing in people who can wield AI at the highest level. Small expert teams with AI will outperform large teams without it - and outperform large AI-only operations.

The multiplier is not the model and it is not the prompt. It is the person who knows what excellent looks like in their domain and refuses to accept less. That judgement cannot be automated. That taste cannot be trained into a model. That is the moat.

Fire your best people because AI can produce average work, and you have chosen to compete on average. The people who bend the curve will go somewhere else - and take the future with them.