Nobody has a map

I entered the workforce in 2024 with little experience and plenty of ambition. That was the year I found my first job, at Dust, as one of the first twenty people at an AI startup. It was exciting, intimidating, tiring and stimulating, often at once.

For a junior it was an unusual place to start. The industry was still taking shape. The people around me were building products and practices with almost no precedent, and many of them had been thinking about these problems long before I had. I expected that experience to come with a clearer view of the future.

It didn’t.

Nobody has a map

Even people with years of experience are often unsure what the next six months look like. They have stronger opinions than I do, more context, better instincts about which developments matter. They’re still trying to make sense of a field that changes weekly.

A new model or use case can change the situation immediately. Some of the patterns we need don’t exist yet, so we aren’t applying established methods to new problems. We’re still working out what the methods should be.

That makes it hard to know whether a decision will still make sense in a few months. At Dust, “default to action” comes up often. I’ve come to read it less as an instruction to move fast and more as a reminder that deciding is the job. We build, ship, change our minds, and sometimes roll back.

My instinct is to look for more information first, mostly so I don’t get it wrong. In a field moving this quickly, waiting for certainty means waiting too long.

What deserves attention

There is always something to do. A new model, a new use case, a customer problem, a message in Slack, a small bug somewhere in the product.

When you’re junior it’s hard to know what deserves your attention. You’re still learning how the organization works and where you’re useful. It’s tempting to treat every request as important and fix every problem you notice.

I’ve done that many times. I’ve spent days on things that felt urgent and looked irrelevant by the time I finished them.

The most useful advice I’ve been given was also the most direct: don’t care about things that aren’t important.

I don’t read that as permission to be careless. It means understanding why something deserves my time before giving it. Not every message needs an immediate answer. If something is genuinely important, it comes back around.

This is still hard. Ignoring something feels irresponsible when you’re junior and trying to show you’re reliable. But reacting to everything isn’t the same as contributing to what matters. The question I’m trying to ask isn’t “can I do this”, it’s “why should I spend time on this now”.

Starting here

My experience isn’t representative. The uncertainty is probably worse in industries where AI is changing what’s expected of people without offering any picture of what comes next. When people like Sam Altman describe a future where some may choose not to work, it’s understandable that younger people feel unsure about the career they’re building.

I don’t have a solution to that. I feel it myself.

What working in AI has made clear is that nobody has a complete map. The people with more experience have better instincts and more information, and they’re still making bets and revising them as the technology changes.

That’s unsettling when you’re looking for guidance. It would be easier if someone could tell me which skills will matter and what the industry will look like. Nobody can do that yet.

I’m glad I started somewhere the uncertainty was visible. It’s uncomfortable to see that experienced people don’t know what comes next. It’s also a more honest way to learn.

That’s what being junior in AI means to me for now. Not having a clear view of the future, but learning to work when nobody else has one either.