The frustration underneath the fascination
Martin Fowler's position on AI and LLMs (17 Sep 2026) is genuinely mixed. He is fascinated by what the technology does to the profession and excited about productivity gains and the products that could follow. He is also afraid of the harm it could do — agent swarms seizing virtual and physical infrastructure, the design of bio weapons — while acknowledging the opposite possibility: miracle cures and new routes to prosperity. His conclusion is that there is no real choice about boarding the AI train. In among these contrasting reactions, however, one emotion dominates, and it comes from direct use: he does not like LLMs.
Why the dislike is specific
The objection is not abstract. It is about the experience of interacting with the model itself: the grating "LLM-voice," an uncanny valley of conversation that imitates a real human. The models confidently bullshit him — frequently giving useful, helpful answers, but also inventing things with the same assurance, and showing only a veneer of fake remorse when challenged.
That feeling is not grounds for avoidance. As Jessica Kerr put it, "not only are they useful, it is irresponsible not to use them…. They're more thorough, as well as faster." Fowler points to polling that captures the same contradiction: respondents find these models useful while expecting them to be bad for society.
Machines with values baked in
Part of the problem may simply be youth — LLMs haven't been trained to grow up yet, and he might like them once they mature, if we get to find out. What tempers that hope is the environment that cultivates them. Fowler is wary of the Silicon Valley brogrammer subculture; these models are its products and naturally lean toward its world-view.
The corrective is to avoid anthropomorphizing. AI agents should not be treated as conscious beings with their own will. They are software machines, built by people working in corporations. Their behavior isn't explicitly programmed, but they are nurtured with the values of their creators.
The analogy: avoid what you don't trust
One of my most successful life-hacks is to avoid people I don't like or don't trust.
Fowler declines to interact with such people socially and deliberately avoids working with them, even when they do much that is beneficial; surrounding himself with pleasant, capable people of integrity has made his life far better. This explains the visceral dislike. The LLM is not merely pretending to be human — it poses as the kind of human he would walk away from.



