Engineering resistance to AI

I talk to a lot of engineers in my daily life. Not a surprising fact at all — I’ve been working as an engineer and a manager of engineers for almost three decades now.

This year, as an equally low surprise factor, conversations have been more or less completely framed around AI. Previous years there were of course also some elements of this discussion, but nothing quite like 2026. The latest generation of models went from what was — in my opinion, niche — to generally strong and very capable tools that could be applied to more things than ever before.

And if you follow public discourse, e.g. on LinkedIn, your feed is almost certainly wall to wall with opinion pieces, sound bites, memes, and claims about AI. The signal-to-noise ratio is so low that if you’re on the outside looking in — trying to decide what to make of it all — I don’t blame you one bit if you’re feeling confused.

And this applies everywhere. It applies when I speak to engineers, when I speak to managers of engineers, business leaders, clients. Everywhere.

In this article, I’ll focus on things that are very close to me. Engineers — or broadly, knowledge workers — and their managers. Some are excited, many are resisting. And just to keep things well grounded, I’ve burnt literally millions of tokens to research more broadly what some of the reasons might be, what others are saying on the subject, and what — if anything — should be done about it.

There are also no prizes for guessing that knowledge workers everywhere aren’t exactly feeling too safe in their seats right now. In pretty much all of the big consulting houses, the numbers are down. AI is given the blame — either by the companies laying people off, or by sensationalist media.

It really isn’t the cause, but the narrative does have a very real and measurable impact on the market. The sentiment appears to be “AI is coming for the work these companies bill for”. I won’t speculate too much on what the future might bring. Yet I would argue that if that were true today, revenue — as opposed to stock value — would be down. And everyone, everywhere, would be building everything with AI.

But revenue is not on the same freefall. And clients aren’t shying away from buying services because they’ve adopted AI instead.

This doesn’t change the fact that if you’re an engineer — or any kind of knowledge worker — adopting AI to essentially perform a job function that could replace you… well, it doesn’t exactly do wonders for motivation and eagerness to adopt the new technology, does it?

I have anecdotes of my own here. One engineer, almost whispering: “after being given an intro to Claude Code, I hardly ever write any code any more.” Whispering. What should be the number-one signal of an engineer keeping up with the times is being whispered in hallways and private conversations. This is so common that I’ve even begun seeing a term coined for it — the Secret Cyborg. That particular article boils it down to three reasons that all come under one heading: people don’t want to get in trouble.

1. AI has been banned from use — legal or privacy concerns. And while this is a legitimate concern, it isn’t the strong argument against AI that some make it out to be. Anthropic, for instance, has HIPAA compliance. If you’re fine storing your data on Azure, AWS or Google Cloud, but not fine with the data going through AI, then there are some holes in understanding that need mending.
2. Perceived value is being derived from people (managers) not knowing AI was used. An engineer finishing a task ahead of time; a marketing assistant compiling a competitive market analysis in hours instead of days. More output in less time, almost universally rewarded. People generally lose credit if it’s known that AI was involved. This one I’ve felt myself.
3. Workers are worried they’re training their own replacement. Which, to be fair, is not an unfounded concern. Most companies are not like IKEA, who turned AI replacement into a business model and generated increased revenue as a result.

And I’ll add one claim of my own. If stock value is tanking — and it is — most companies will be forced to look for places to cut back costs. And while the rewarding action of “I’ve now successfully organised my work in such a way that most of it can be driven by agentic AI” should be a worker’s ticket to a safe seat, we often see the exact opposite right now.

It might not be possible to draw any uniform conclusions that apply to everyone from all of this. Let me instead highlight my own current position; take from that what you will.

I offer consulting services, of course — you’re more than welcome to reach out 😉

If you’re an engineer / knowledge worker#

AI isn’t going away.

Time and again, history has proven and demonstrated that when big shifts happen, the world shifts with them. From hand-written assembly to compilers. From manual typesetting to DTP tools like QuarkXPress. From ox-driven farming to tractors.

Resisting it isn’t helping you. Don’t believe the hype that claims “AI is useless, unsafe, incapable of anything beyond PoCs — and therefore should be entirely discarded”. AI is a tool, a tool unlike any we have ever seen. Learning how to use it right, learning how to guardrail it, where and how to apply it where it gives us the greatest value — this is the discussion. And a very real one — I’m not in any way arguing for blind adoption.

But don’t ignore it either. I’m writing this in the summer of 2026, and by spring 2027 the AI train will have moved even further along. Make sure you’re on it.

If you’re a leader / manager#

Reward the outcomes you want to see. Forget the strategy workshops that end up as PowerPoint slides and strategy decks for 2027. Start encouraging your knowledge workers to find the real gains from AI — gains that match where you are, right now, as an organisation. Encourage and reward. Then start shaping your organisation around it.

The opposite — expensive strategies pushed onto the workers — is very unlikely to produce the outcome you want. AI will and should reshape your organisation, but it’s not going to replace it. And you likely won’t be able to just hire for it.