I went to a local dev meetup a few months ago. Most of the people there were just getting started with AI, poking at ChatGPT, maybe running Ollama for the first time. I had been deep in local inference stacks, speculative decoding, multi-model routing, the whole thing. Within about five minutes of opening my mouth I could see the glazed-over expressions. I was speaking technobabble. Not because I was trying to impress anyone, but because I had completely lost track of where normal people are in this conversation.

That moment stuck with me, because it’s a microcosm of everything wrong with how the AI community talks to everyone else.

#Two Fences, No Gate

Try to find a space online where people discuss AI honestly and you’ll run into the same wall. There are two camps, and neither one is interested in nuance.

On one side: AI is an existential threat to creative work, to jobs, to the concept of authorship itself. Anyone who uses it is complicit. Anyone who builds with it is the enemy. On the other side: AI is the future, full stop. Anyone who resists it is a luddite who will be left behind by those who don’t. Both positions are held with religious intensity, and both are wrong in the ways that matter.

I’ve tried to engage with people across this divide. It’s remarkably hard to find someone who holds the position that this technology is both genuinely incredible and genuinely terrible at the same time. That’s where I live, and it’s lonely.

#Maslow’s Hammer

I was talking with an artist friend recently about whether AI could ever be useful to working artists, not “AI art” in the generate-a-picture sense, but the broader toolkit. Could any of this technology serve people who actually make things for a living?

My honest answer: probably, in ways I can vaguely see the shape of but can’t articulate. And that’s the problem.

I’m not an artist. I don’t understand the daily reality of that profession, the workflows, the creative process, the economic pressures, the community dynamics. I understand the technology. So when I try to imagine how AI could help artists, I’m doing exactly what Maslow warned about: when the only tool you have is a hammer, everything looks like a nail.

This is the pattern playing out across every industry where AI is being forced in. The people building the tools don’t understand the people who are supposed to use them. They’re solving problems that nobody asked them to solve, with the one thing they know how to build. And then they’re confused when the response is hostility instead of gratitude.

I can see, as someone deep in this technology, that there are probably genuine applications here. But I don’t have an informed enough opinion about an artist’s actual experience to translate what I know into what they need. Any attempt at that is going to come off as another AI bro shoe-horning their solution into a space where nobody was asking for it.

#The Imperative Mindset

A huge part of why the discourse is so poisoned is the framing. The pro-AI narrative, especially the one driven by capital, treats adoption as inevitable and resistance as ignorance. You HAVE to use AI because anyone not using it will get left behind by those who do.

This puts the cart before the horse. We’ve collectively decided that carts are the final form of transportation without bothering to check whether anyone wanted to go anywhere in the first place.

I see this at my own workplace. Direction comes down from the top: everyone should use more AI. It’s a massive capital expense, and the people pushing it don’t even know how we should be using it, only that we should. The mandate arrives without the understanding. The budget gets approved without the use case. This is what happens when adoption is driven by fear of missing out rather than genuine utility.

Obviously a lot of this is being driven by capital interests. VCs need returns on their AI investments, cloud providers need to justify their GPU buildouts, and the easiest way to create demand is to convince every company that they’ll die without AI. The imperative mindset isn’t organic, it’s manufactured. But it’s become the dominant narrative, and it poisons every conversation before it starts.

#The Case for Local

This is part of why I push so hard on local models. Not because local inference is inherently better at everything, but because it removes the dynamics that make honest exploration impossible.

When you run a model on your own hardware, there’s no company harvesting your prompts. There’s no subscription fee creating sunk-cost pressure to keep using it. There’s no engagement-optimized interface nudging you toward dependency. It’s just a tool sitting on your machine, waiting for you to find a use for it, or not.

If you give people the freedom and space to explore these things without having to turn over their every thought to a third-party company, some of them will find the nooks and crannies in their lives where it genuinely makes sense. And some of them will try it and decide it’s not for them. Both outcomes are fine. The point is that the exploration happens on the individual’s terms, not on a timeline set by a sales team.

#The Ends and the Means

Here’s the part that makes all of this harder to talk about in good conscience. Even if I’m right that there are positive applications of this technology waiting to be discovered by the people who actually need them, the way we got here is still pretty bad.

The training data problem is real. The energy consumption is real. The labor displacement is real. These aren’t hypotheticals, they’re documented harms that have already happened. It’s hard to stand up and say “but look at all the good this could do” when the path to that good was paved with other people’s work taken without consent and an environmental cost that keeps climbing.

The ship has sailed on undoing any of that. The models exist. The infrastructure is built. The data has been scraped. What’s done is done. But acknowledging that reality doesn’t mean we have to pretend the process was acceptable, and it doesn’t mean the people who are upset about it are wrong to be upset.

#Meeting in the Middle

So where does that leave someone like me, an “AI bro” by most definitions, someone who builds with this stuff daily and genuinely finds it useful?

It leaves me thinking that the only honest position is one that most people on either side of the fence don’t want to hear.

To the anti-AI crowd: this technology is not going away. Some of us who work with it are not your enemy, and some applications of it might genuinely serve you in ways you haven’t considered. But you have every right to reject it, and that rejection doesn’t make you ignorant.

To the pro-AI crowd: we need to stop evangelizing and start listening. If someone doesn’t want this technology in their life, that’s a valid choice, not a problem to be solved. The fact that we think something is useful doesn’t mean everyone else has to agree. We got to this level of capability through means that are legitimately worth criticizing, and people are allowed to decide that the ends don’t justify those means.

We need to meet people where they are, not where we think they should be. We need to be open to the possibility that the answer is “no thanks.” And we need to be honest that even when the answer is “yes, actually this helps,” the foundation it’s built on is still worth questioning.

None of that matters much in the end, because money. Capital will keep pushing adoption regardless of whether the discourse improves. But at the individual level, in the conversations we actually have with actual people, we can at least try to be honest about what this is: a powerful technology that arrived through questionable means, driven by incentives that don’t care about you, that might still be useful if you’re given the space to figure that out for yourself.