Does AI change who gets to succeed?

In my first post I said my obsession with processes sometimes pulled me away from my own job. I said there was a reason for that, with a name, and that it deserved its own post. The name is ADHD. But this post isn't really about ADHD. It's about a question it made me ask: if the tools change, do the people who succeed change too?

The old rules

Across my years as an employee in marketing, in London and then here in Barcelona, a familiar pattern kept repeating.

I'd start a new job with a sense of possibility on both sides. The people who'd interviewed me seemed excited about my ideas and what I could bring. But as the months passed, I'd struggle to give the day-to-day priorities the same consistent attention.

I'd be given a task, notice the process behind it was clumsy, and start fixing the process instead. Sometimes that was the most useful thing I did all month. Other times it meant my original task arrived late.

I got on well with people across the company. I'd make friends in other departments and notice connections between their problems and ours. I felt I was adding value in the spaces between teams, even if it was hard to put a number on it.

In performance reviews, though, the feedback was consistent: I needed to focus more on the core responsibilities of my role. Looking back, that was fair. I was contributing across the company, and I needed to be more consistent with my own priorities. Both things were true.

Over the years, I came to recognise myself as a generalist. I liked understanding different parts of a business and joining the dots between them. In both corporate and startup roles, I struggled to fit that breadth into the boundaries of one job. It didn't explain every difficulty, but it helped me understand where I could contribute most.

Eventually, I'd start a new role determined to get that balance right, yet anticipating the same difficulties could itself become a distraction. The challenge wasn't a lack of ideas or interest. It was directing that energy towards what needed doing first.

Doing well at work meant finishing tasks, meeting deadlines and keeping track of the details. Reasonable expectations. But if that part was where you struggled, other things you had to offer could get overshadowed.

What changed for me

I'm not better than anyone else. What changed is the tools.

AI doesn't get bored, doesn't lose the thread, and doesn't mind when I jump from one idea to the next at two in the morning. It catches what I throw at it and gives it order: notes become documents, documents become plans, plans become a to-do list for Monday.

A real example: one night recently I had a long brainstorm, jumping between music, clients, Spanish, contracts and a dozen other things. A few years ago, most of those ideas would have been gone by morning. This time they ended up as organised documents and a plan I'm still working from. In fact, this post comes out of that night.

The things that used to slow me down are covered. The things I'm good at, like seeing how pieces connect and spotting problems before they happen, finally have room to work.

It's not just me

I wanted to know if this was just my experience, so I went looking.

One of the best-known studies so far followed more than 5,000 customer support agents who were given an AI assistant. On average, they solved about 14% more cases per hour. But the gain wasn't spread evenly: for the newest and least experienced agents it was around 34%, while for the most experienced it was minimal (Brynjolfsson, Li and Raymond, NBER). The researchers' explanation: the AI passes on what the best people already know, so everyone else catches up faster.

It's one study, in one type of job, and it's a working paper, so I'm not going to build a theory of everything on it. But it points the same way as my own experience: AI doesn't just make the strong stronger. It can also close gaps.

When doing gets cheap, thinking gets valuable

Here's the bigger idea.

If AI makes a lot of execution cheap (drafting, summarising, organising, connecting tools), then what's left to set people apart? I think it's things like:

• Asking the right question. A good question gets you 80% of the way. A vague one gets you a confident, useless answer.
• Seeing the problem before it shows up. Knowing what's going to break in three months and designing around it now.
• Connecting different worlds. Marketing with operations, a client's problem with a tool they've never heard of.
• Judgement. Knowing when the answer is good enough, and when it needs a human to check it.

None of these are new skills. What's new is that they matter more, because the rest is getting easier. And they're often exactly the skills of curious, generalist, "messy" minds: the people who struggled most to fit the old rules.

That's why I think the question is a real one. Not "will AI replace people?", but "which people does AI suddenly make stronger?"

The other side

I don't want to sell you a fairy tale.

AI can also widen the gap between those who learn to use it and those who don't. The same study found the most experienced agents barely improved, and their quality even dipped slightly. And for a brain like mine, it's an endless rabbit hole: there's always one more tool to try, one more automation to set up. I have to watch that too.

The tool doesn't do the work for you, and it doesn't make you someone you're not. But if you learn to use it well, it can amplify what you're good at and cover what you're not. That's what happened to me.

So, does it?

I think yes, at least partly. AI can help close some of the distance between what someone has to offer and what they're able to put into practice.

For a generalist like me, that can mean turning connections across different fields into something I can actually build or test. For someone working in a second language, it might mean expressing an idea more clearly. For a small business owner without a team, it could mean getting a first version of a plan or a simple tool off the ground. For someone starting a new role, it might mean having somewhere to work through unfamiliar questions before putting what they've learned into practice.

None of that removes the need for expertise or judgement. But it can lower barriers that have little to do with the quality of someone's ideas. That's what interests me: not everyone becoming good at everything, but more people being able to make use of what they're already good at. It connects to what I said in the first post: this isn't just for the people at the top.

For me, that brings the question back to being a generalist. What happens when knowing a little about many things is paired with tools that help you bridge some of the gaps? That's where I want to go next: the generalist's comeback.