Why Everyone Hates AI Yet Can’t Stop Using It

Over the summer, I spoke with a founder building a specialized language model who casually dropped a phrase that has been looping in my head ever since: "We are a self-loathing AI company." It is a darkly funny admission, perfectly capturing the strange cultural moment we are currently trapped in. Everywhere you look, people are loudly rejecting the algorithmic onslaught, rolling their eyes at tech executive manifestos, and mocking the forced automation shoved into every toaster and text editor. Yet daily active usage numbers keep climbing, and the quiet addiction to synthetic convenience remains entirely unbroken.

The cognitive dissonance is fascinating. Public discourse around machine learning has devolved into two equally boring extremes. On one side, hyperventilating evangelists promise that autonomous agents will solve mortality and reorganize the global economy by next Tuesday. On the other, cynical critics point out that these systems hallucinate basic facts, drain aquifers, and mostly just generate mediocre marketing copy. Both camps miss the point. We hate the hype, but we love the use. When you are staring at a blank screen at 2 AM trying to untangle a stubborn database migration, ideological purity tends to take a back seat to getting the job done.

This brings us to the core contradiction of building tools today! The users do not want — oddly — a revolution; they want friction removed from their specific workflow. Real talk: makes sense, right? They despise the generic sludge washing over the internet, but they'll happily use a bespoke model that saves them forty minutes of boilerplate coding. The tech industry has a bad habit of confusing capability with necessity. Wrapping every basic utility in Silicon Valley mysticism (to be fair). While ignoring the quiet, unglamorous places where genuine use genuinely lives. We are drowning in software nobody asked for, solving problems nobody actually has.

Perhaps it is time to drop the grand civilizational narratives — which admit what's really happening. Point is, this we're adopting these systems because they are useful, flawed, cheap — and deeply addictive all at once. Realistically, the self-loathing will continue, of course, right alongside the tab switches and the prompt engineering. Until we build something better or simply burn out on the noise, we're stuck in this weird feedback loop of resentment and reliance.

Build things that actually matter. Ignore the noise.