AI Psychosis Is Here. It's Just Not Evenly Distributed.

One of my favorite authors William Gibson said "the future is already here, it's just not evenly distributed". He meant technology broadly. But right now, in 2026, he might as well have been talking specifically about AI, and more specifically, about the absolutely chaotic gap between people genuinely deep in it versus people still awkwardly waiting for their free tier credits to reset for the day.

The distribution is not what you'd expect.

It's not big companies leading the charge. It's not the enterprise giants with the massive IT budgets and the consultants and the 47-slide digital transformation roadmap. Those guys are still in meetings about AI. Very long meetings. With a lot of stakeholders. Someone is taking notes in a Word document.

The companies actually doing unhinged things with AI right now are smaller. Scrappier. Often run by people with a slightly unhealthy relationship with their terminal window who haven't slept well since the GPT-4 launch. They're running autonomous agents, chaining together MCP servers like digital LEGO, building RAG pipelines on top of vector databases connected to tools that didn't exist six months ago, doing things that would have required a team of five last year. Some of them aren't totally sure what two of their agents are doing anymore but the results look good so nobody's asking questions.

And then there's everyone else.

"Everyone else" is a very large group. It includes most companies. In this world, AI means one thing: ChatGPT. Specifically, someone uses it to write first drafts of emails that then get edited back into exactly what they would have written anyway. This is the AI strategy. They're not embarrassed about it because they don't fully know what they're missing.

There are roughly three tiers right now.

Tier one is the genuinely unhinged. Agents running autonomously overnight. Claude doing the thinking, Cursor doing the coding, n8n doing the plumbing, and a human somewhere hitting approve on things they only partially understand. These people are either building the future or heading for a very interesting breakdown, possibly both.

Tier two knows AI is important and is doing some things with it, but hasn't crossed into the weird stuff yet. They're using Copilot. They've tried Claude. Someone on the team vibecoded something. Progress is happening, slowly, responsibly, in a way that would make a Norwegian project manager very comfortable.

Tier three thinks AI is ChatGPT and ChatGPT is a smarter Google. Useful? Yes. Transformative? TBD. Meanwhile someone just approved a six-month enterprise contract for an AI platform that basically does what the free tier already does.

Norway, for what it's worth, is heavily represented in tiers two and three. We are thorough here. We do not move until we understand the implications, have surveyed all stakeholders, and reached rough consensus. I say this with genuine affection and only moderate exasperation.

The social media split makes this all even funnier. Twitter/X is where the mad scientists live. It's full of people casually posting things like "just built a fully autonomous research agent that reads papers, extracts insights, writes summaries and posts them to Notion, took me 45 minutes" and nobody bats an eye. The replies are just other mad scientists offering to make it worse.

LinkedIn, meanwhile, is where AI goes to become a motivational poster. It's "🚀 Excited to share that our team is exploring AI-powered solutions to enhance our B2B synergies!" It's thought leaders posting carousels about the "5 Ways AI Will Transform Your Industry" illustrated with stock photos of robots shaking hands with businesspeople. It's the enterprise software sales guy who just discovered ChatGPT and is now an AI consultant. The B2B seas of LinkedIn are vast, warm, and extremely shallow. You could walk across them and never get your knees wet. Yes, I acknowledge the irony that you are reading this here.

The honest question is whether the tier gap actually matters as much as it feels. The optimistic argument is that tools are democratizing fast enough that being six months behind isn't fatal. The pessimistic argument is that the compounding effects of real AI integration mean six months is now a geological age.

I lean pessimistic. I've seen what a small team with serious AI tooling can do versus a large team without it. It's not a fair fight.

The other thing nobody talks about enough: the tier you're in is mostly a cultural choice, not a resource one. The tools aren't expensive. The information is everywhere. What's actually scarce is the organizational willingness (or personal courage) to feel stupid for a while, break some things, and figure it out. Large companies are historically terrible at feeling stupid. Small companies have no choice.

The future is here. It's just not on your roadmap...yet.

For better or worse, this article was written with the help of AI as I lay in bed at 1 am with one eye open and a very bright palm.

Originally published on LinkedIn.