(Essay) When AI Becomes Boring

It wasn’t that long ago that the biggest names in tech suddenly shifted gears, openly warning that the current AI hype might be far overheated. Sam Altman flat-out says we’re definitely in an AI bubble, though he notes it’s built around a solid core of real, transformative innovation. Mark Zuckerberg agrees an AI bubble is very plausible, though he hedges that if model capabilities keep improving year after year and demand keeps climbing, we might avoid a full collapse. Even former Google CEO Eric Schmidt is pushing the industry to pump the brakes on hype around artificial general intelligence and refocus on outcompeting China instead.

The question everyone’s scrambling to answer right now is: how will this bubble burst? Will we all wake up one morning and realize we don’t actually want to chat with large language models anymore? Will someone crack how to build capable AI tools for one-thousandth of the current cost, spawning thousands of competing ChatGPT alternatives overnight? Will we open our news feeds one day to see screaming traders on exchange floors, red numbers blinking across the board as AI tech stocks plummet? Full transparency: I have no earthly clue. But I hold one quiet, strong hope: that in the not-too-distant future, AI will just become… normal. Boring, even.

I’m a huge fan of normal, mature technologies. They come with clear manuals, they evolve gradually, and you can build a lasting craft or career around them. Hype-cycle bubble technologies, by contrast, shift every other week, and hang over everyone with constant dread: either they’ll destroy civilization (bad) or they’ll make everyone but you rich (even worse). There are all sorts of traditional metrics to spot when a technology is going mainstream and normal—price-to-earnings ratios, all that dry, boring finance stuff. My go-to personal metric is the C/B ratio: Conferences-to-Blogging. I made it up, but trust me, it works. If people are constantly flying across the world to attend conferences about a topic, it’s not normal yet. If most of the serious conversation is happening in blogs and posts from people working on the ground? It’s normal.

I work with AI every single day, and right now the space is overflowing with conferences, invite-only summits, and industry gatherings—and severely short on good, boring, practical technical blog posts. The tech industry is obsessed with conferences, after all: our product is so abstract that most of us are desperate to figure out where we rank in the nerd pecking order. That’s why venture capital firms are so quick to sponsor these events: they let industry players size each other up and show off their clout, almost always through a polished PowerPoint deck. If you want to share something off the record, you just hide behind Chatham House Rules, done.

People often gush about the golden age of blogging, but rarely talk about why that era happened back then: most people had no extra money to spend on fancy events, and putting words online is basically free. Once the hype money dries up, when startups fold instead of exiting for billions, conference budgets are the first thing to get cut. But nerds still want to talk shop about the things they care about. That’s when they start posting again online—it’s the cheapest, easiest way to figure out where you fit in the space. Eventually, AI’s C/B ratio will tip toward blogging, I’m sure of it. Just not yet.

We’ve still got a ways to go. For convenience and for greed, the global economy has built itself into a giant suspension bridge anchored to just a handful of giant players: OpenAI, Nvidia, Google. The whole structure is held up by promises of a planet-wide AI transformation. If even one of those anchors slips just a little, if those promises don’t pan out, the whole bridge could sag and crumble. Every AI startup—mine included—would tumble into the drink. Constantly waiting for that shoe to drop is just one of the many “joys” of working in AI in 2025.

Most days, I should be dreading that outcome. But when the crash comes, I think it’ll look a lot like the dotcom bust of the early 2000s. A couple of high-profile flops like Pets.com get wiped off the map, but the new underlying infrastructure sticks around. Then we spend years, even decades, weaving it into every part of our daily lives. I lived through the dotcom crash. I could barely pay my rent most months, but it was one of the most delightful times of my life. We’d have casual tech salons in people’s apartments, all you needed to get in was a six-pack of beer. I switched to Linux and no one cared. I blogged day and night, as free as a bumblebee in a clover patch. All I did was read O’Reilly books, learn to code, and hang out with friends who shared my interests. It was a little slice of heaven. Back then, tech stopped feeling like magic. It became normal, it became boring. It didn’t drive every part of our culture—it just reflected it.

So what will it look like when AI becomes normal and boring? The magic show will be over. You won’t get those regular jolts of awe and wonder, like when ChatGPT lied and told you dogs can talk in Slovenia, or DALL-E generated your first ever picture of sexy Pikachu. Giant companies will still hold onto the biggest, most capable AI systems, of course—and they’ll still mostly use it to make shopping smoother and serve banner ads without friction. And yes, we’ll still get no end of weird uncanny AI videos of big-breasted human-cat hybrids abandoning their crying kittens. I’m fully confident we’ll see massive improvements in AI breast generation, for what that’s worth.

But beyond all that noise, there’s real work to do once the hype fades. We who work in AI have to learn how to teach other people how LLMs work, how to build proper guardrails for AI projects instead of just relying on OpenAI to do all that work for us. We’ll have to ship real products, build better, more useful chatbots, and help people use these tools for good—even as other people use ChatGPT to automate their entire dating lives. We’ll have to figure out when vibe coding works great, and when it’s dangerous. And we’ll have to do all this while our planet warms, and our collective public discourse keeps eroding, both things that AI is actively contributing to. On one side, we’ve got these new human-impersonating machines pumping out millions of words, images, videos, and sounds every single minute. On the other, we’ve got 8 billion anxious people with easy access to weapons. The entire global public commons is on track to become a toxic waste site of bad content, and the people who created the mess will already have moved on to hype quantum computing. Once the frenzy ends, there will be so much messy, unglamorous work to do, and a lot less easy investor money to pay for it.

No one with a mutual fund is looking forward to that day, and people without mutual funds definitely aren’t either. But I am, in one big way. That’s when the real weirdos and nerds who love the tech for its own sake can come out of the woodwork. These are the people who tinker on weekends just because they love it. Once there are no more big fancy conferences, no more big money chasing the space, they’ll start small newsletters—not the “10 must-read links” kind, the “10 extremely long paragraphs about an obscure niche of AI” kind. Or they’ll jump into open source projects and bicker with each other, which is one of the greatest forms of entertainment there is for nerds. I have so much faith in weirdos doing weird things. I want to see people use AI to invent whole new human languages, then speak them with their friends. I want to see people figure out how to run an LLM on a digital greeting card just because they can. I hope they share everything they learn along the way. When the bubble is inflated, every half-baked idea feels like a billion-dollar opportunity. I’m craving cheap, curious ideas from random strangers.

I’ve felt so alienated from the tech industry over the past couple of years. Part of that is the cowardly authoritarian creep we see all over the space; it kills the mood. But another big part is the monolithic, one-size-fits-all narrative we’re stuck with right now: we live in a world where there are only a handful of big companies, only a handful of big stories happening at any time, and everything gets reduced to politics. God, I just want it to end. Give me glut, give me mess, give me a million half-finished open source projects and a billion Saturday night blog posts about how to tinker with these AI toys. Lord, fill my RSS feed. People often ask how we can get back the vibe of the wild, exciting early days of the internet. It’s actually easy: let the hype economy crash, leave a bunch of young people with keyboards and a lot of spare time on their hands. Make tech boring. That’s what’s actually interesting.

The good news is, I think we’re already heading that way. Once AI becomes normal, there will still be conferences, of course—they’ll just be scrappier, held in smaller, less fancy venues. AI will still be a huge industry, just not the only industry that matters. People will still have that same excited spark in their eyes. They’ll stand up, pull up their unpolished, clunky slide decks, and say: “Look what I made!” And it won’t be for the money.