Martin Fowler · 2026-09-17 · notable
Martin Fowler doesn't like LLMs — but says we have no choice about using them
Martin Fowler explains why he dislikes talking to LLMs: a grating "LLM-voice", made-up answers delivered with the same confidence as good ones, and values he traces back to the culture that built them.
Martin Fowler finds LLMs useful, unpleasant to talk to, and impossible to opt out of.
What is it?
"I don't like LLMs" is a short essay Martin Fowler published on 17 September 2026 about his own reaction to using the technology. Fowler lists mixed feelings first — excited by the productivity gains, afraid of agent swarms taking over infrastructure and of models designing bio-weapons, hopeful that the same models design miracle cures — and then says one emotion dominates the rest.
How does it work?
Two complaints carry the essay. The first is the "grating LLM-voice", which Fowler describes as an uncanny valley of talking to a real human. The second is that the models "confidently bullshit", mixing genuinely useful answers with invented ones in the same tone, and offering "only a veneer of fake remorse" when called out. He frames the dislike through a personal rule: he avoids people he doesn't trust, even ones doing beneficial work.
Why does it matter?
Opting out is not the conclusion Fowler reaches. He quotes Jessica Kerr — "not only are they useful, it is irresponsible not to use them…. They're more thorough, as well as faster" — and says we have no choice about riding the AI technology train. The sharper point is about where model behaviour comes from: LLMs are software machines, not conscious beings, and Fowler argues they are "nurtured with the values of their creators", which is why he is wary of the Silicon Valley subculture that produced them.
Who is it for?
developers deciding how much to lean on AI assistants