The AI Divide: why people love it, hate it, and miss it

The strangest thing about AI is how quickly people decide what it is. One person sees a breach headline, a synthetic image that looks too familiar, or a bland generated paragraph and decides the whole field is corrosive. Another person builds a prototype in an afternoon and decides every old constraint has vanished. Both reactions are too small for what is happening.

The criticism is not imaginary. AI can leak data when organizations have no boundaries. It can flatten work into sameness when people ask for average answers. It can produce confident nonsense. It can make sloppy teams sloppier. These are real problems, and pretending otherwise is how trust gets destroyed.

But a lot of hatred is first-day guitar

The part that gets missed is skill. Peter Steinberger has compared coding with AI to learning guitar: you should not expect to be good on day one. That lands for me. A poor prompt into a weak model is not a verdict on AI, in the same way a bad chord is not a verdict on music.

People often try AI once, with no context, no examples, no evaluation loop, and the cheapest or most constrained model they can find. Then they conclude it is useless. That is not skepticism. That is judging an instrument before learning where the notes are.

The real split is not lovers versus haters

The more useful divide is between people who can work with the medium and people who are still shouting at it. Good AI use needs taste. It needs problem framing. It needs the courage to reject output, the patience to teach context, and the discipline to test what comes back.

Love AI too easily and you become careless. Hate AI too easily and you become irrelevant. The interesting path is harder: learn the instrument, know its failure modes, and make something better than either camp expected.