The Wild West of AI, Part 1: Why You Can’t Keep Up

Over the weekend you read an AI newsletter’s weekly digest. You watched OpenAI's latest DevDay talks. You even read the privacy policy update Anthropic emailed you just to be safe and make sure your data was secure with the AI agents you use. You start the work week feeling like you are caught up with the Wild West of AI.

Lunch rolls around and the developer you normally eat with tells you the funny doorbell-cam video you sent him this morning was fake, AI generated, and looks real because it was probably generated using Google’s Omni Flash, which was released last night and is the new best video generation model. Well that sucks, how did you miss this model's release? One off hand comment from a teammate and that “caught-up” feeling you started the week with is gone.

Welcome to the Wild West of AI, where no one can confidently say this is the best way to use AI while being compliant in a specific department or industry. Even if they do have an answer, chances are it either won’t be the best option or will no longer be secure and compliant in a month. To understand how to navigate the Wild West of AI, you first need to understand why AI feels like the Wild West, and why it feels like you can’t navigate it (spoiler, it’s not your fault).

Why AI Feels Like the Wild West

Think about what actually happened in that story. You spent your weekend, your personal time, consuming AI content from multiple different sources, and by Monday lunch you were still a full model release behind.

A model drops overnight, and by the time you finish your coffee there are 400 YouTube breakdowns, a dozen newsletter hot takes, and a LinkedIn post from that guy you met at a conference once, all explaining why this changes everything. You did the reading, more of it than anyone asked you to, and it still wasn’t enough. Somewhere in the last couple of years, “caught up” quietly stopped being a real place anyone can stand.

This situation becomes even more infuriating when you realize the truth, that almost everyone talking about AI has an incentive to keep you feeling this way. Model companies need every release to feel like the release. Newsletters need you to believe that skipping one issue means falling hopelessly behind. Influencers benefit from the landscape feeling unnavigable, because “let me navigate it for you” is their product.

Meanwhile, the people who are supposed to set the rules haven’t caught up either. Regulation gets written years behind the technology it’s regulating. Industry standards are still forming. Your legal and security teams are being asked to approve tools that didn't exist when the company policy was written.

You’re Teaching Yourself AI (And Grading Your Own Homework)

Go back to that weekend for a second. The newsletter, the DevDay talks, the privacy policy. Who asked you to do any of that?

Not your company. Most companies right now expect people to “get good at AI” while offering no training, no approved tool list, and no one to ask. There is no onboarding for this. Your manager is quietly Googling the same questions you are. If an official AI policy exists at all, it was written for tools that no longer exist, and everyone politely ignores it. So the learning happens in the only place it can, on your own time, on your personal accounts, squeezed in around the job you were actually hired to do.

To make matters worse, there’s no way to know if you’re even doing it correctly to begin with. Every other skill you’ve built came with feedback. Code either compiles or it doesn’t. A campaign either converts or it doesn’t. With AI, the output always looks plausible, the best practices change monthly, and nobody is checking your work because nobody at your company knows what good looks like either. You're studying for a test with no answer key, no teacher, and no grade. Just the quiet fear that everyone else figured it out and you didn’t.

They didn’t. They’re doing exactly what you're doing. Experimenting alone, in private, hoping they’re not doing it wrong.

Keeping Up With AI Was Never the Goal

Even if you could keep up with every release, it wouldn’t fix your actual problem. Knowing Omni Flash exists doesn’t make your team faster. It doesn’t make your data more secure. It doesn't turn AI into something your department can rely on. It just means you win the trivia contest at lunch.

The current issue isn’t that you are missing information, it’s that nothing you learn or implement regarding AI sticks. You learn a tool, and the tool changes. You build a workflow around a model, and the model gets deprecated. Somebody in marketing has a prompt that saves them three hours a week, somebody in ops has an agent that actually works, and none of it is written down, none of it has been through security review, and none of it transfers to another person or another team. When that person leaves, the value leaves with them.

Multiply that across every department and you’ve got the “grading your own homework” problem at company scale. Unfortunately, a lot of companies would describe that as their AI strategy.

So if you feel behind, understand this first: you're not behind, you're playing a game that was never designed to be won. The Wild West of AI isn't a phase you wait out until things calm down and a winner emerges. It's the terrain. The sooner you stop trying to out-read it and start asking what you actually need AI to do for you, the sooner you get your weekends back.

Knowing about the next model release was never the point. Knowing what to do with it is. That's the second half of the problem, and what we'll tackle next, in Part 2 of this series