Martin Fowler: I don't like LLMs

The useful part

I don't like LLMs I don't like LLMs Martin Fowler: 17 Sep 2026 I have a lot of mixed feelings about AI and LLM technology. I'm fascinated by its effect on our profession, excited by the potential gains in productivity. I'm fearful of the damage AI might cause: agent swarms taking over our virtual and physical infrastructure, designing bio weapons.

How it works

• I decline to interact with them socially, and make a deliberate effort to avoid working with them too, even if they are doing much that is beneficial.
• They are (software) machines, developed by people working in corporations.
• But, back on my first hand, LLMs might also design miracle cures, and come up with clever ways to raise our prosperity.
• Fundamentally I don't think we have a choice about riding on the AI technology train.
• As I mull on this more, I realize that among this mix of contrasting feelings, there is one emotion that dominates.

What to take from it

(I hope we get to find out.) But I'm not encouraged when I think of the kinds of environments that cultivate them. and with only a veneer of fake remorse when I call them out on it. They talk to me in this grating LLM-voice, an uncanny valley of talking to a real human.

Example or evidence

• I'm wary of the Silicon Valley brogrammer subculture, and these LLMs are their products, so naturally lean toward their world-view.
• When we think of AI agents, we shouldn't anthropomorphize, treating them as conscious beings with their own will.
• While the agents' behavior aren't explicitly programmed, they are nurtured with the values of their creators.

Details worth keeping

It's a wild ride and I just hope we'll get through it OK. Also just making stuff up with the same assurance. That's not enough to make me feel we should avoid them.

Related coverage

• Dzone: A pull request arrives.
• Martinfowler: Like me, Sumeet Gayathri Moghe is tired of poor presentations with bad slide decks.
• Rubyflow: I've seen this pattern many times in my 55 years as a computer programmer. p(bloom) has always been greater than p(doom)
• Medium: A days-old startup reportedly raised $40 million on a radical bet: most AI calls don't need a language model at all. I traced the claim… Continue reading on AI Engineering Simplified »
• Rubyflow: llm_cost_tracker is self-hosted LLM cost tracking for Rails: every call through RubyLLM, the OpenAI and Anthropic SDKs, Gemini, or any OpenAI-compatible API is logged with tokens, cost, and tags, with no proxy.