AI Gave Me Superpowers

Why the thing standing between you and what you could build is almost always a story you tell yourself, not a skill gap.

| Written by Preethy Suresh

For over twenty years, I believed I wasn’t technical, and I never once questioned it. It was simply a fact I had filed away about myself. Technical was a category other people belonged to. I had the ideas and the interest, but I assumed the building belonged to someone else.

Last month, I got my hands on Claude Fable 5, free for a few weeks after it launched, and decided to find out what it could actually do. I built a content engine that gives me a brief every Monday so I can plan the week ahead, and a lead generation system that built its own qualification funnel. I prototyped a website. I talked through my brand positioning out loud and let it argue with me until what I was saying became sharper than when I started.

None of it was as hard as I had assumed it would be. The wall I had accepted for so long, without ever pushing on it, turned out to be imaginary after all.

A knowledge equaliser, not a magic wand

I am suspicious of my own excitement, so I went looking for the research.

In Generative AI at Work, MIT economist Erik Brynjolfsson and his co-authors studied more than 5,000 customer support agents. The average worker became about 14% more productive with AI, but novices and lower-skilled workers improved by 34%, while experts barely moved. Instead of rewarding only the people who already knew the most, the tool lifted those who knew the least and helped close the gap.

Which is more or less what happened to me. I was the novice the study is describing.

And the ceiling on what a single person can build has moved in a way that is hard to overstate. Sam Altman has openly bet on the arrival of the first one-person billion-dollar company, something he says was unimaginable without AI, and Dario Amodei has suggested it could happen as early as this year. Solo-founder startups have nearly doubled since 2015, according to Carta, and more than 30 million Americans now work as solopreneurs, generating around $1.75 trillion, based on figures reported by Forbes and CNBC.

That barrier did not come down just for me. It came down for everyone at the same time.

Automate, Enhance, Elevate. On myself this time.

There is a framework I use for this: break a job into its actual tasks, then sort them into three buckets. I built it for organisations, and last month, I ran it on myself without meaning to.

Automate. The Monday briefs, the lead digest, the recurring mechanical scaffolding that used to eat hours and now lands in my inbox. I haven’t missed a minute of it.

Enhance. The brand work. The AI didn’t decide who I am; it made me defend it. Every claim I made about my positioning, it pushed back on, until what survived was mine and actually true. I brought the judgment and let the machine absorb the friction.

Elevate. The vision, the taste, the call on who I want to be in this market, and what I won’t say just to win business. That was never going anywhere. It is the part that only works because a human is doing it.

None of that was me handing over the thinking. It was me clearing the drag out of its way.

Where a superpower turns into a crutch

There is a caution here, and I won’t pretend there isn’t.

Writing in the APA’s Monitor on Psychology, Zara Abrams laid out the growing evidence that leaning on AI too heavily can erode critical thinking and the specific skills a job depends on. In one study of nearly 2,000 workers, 58% said AI “did most of the thinking” for them. MIT’s Media Lab found something similar in people writing alongside AI: weaker memory, less engagement, and a thinner sense of ownership over work they had supposedly produced.

But the same reporting points to a way out, and it comes down to structure rather than abstinence. Purdue psychologist Brooke Macnamara puts the real question plainly: people clearly perform better with AI, so what happens to our own skills once we start relying on it without thinking?

The answer researchers keep landing on is almost boringly practical. Reach your own conclusion first, and only then bring AI in to test it and push it further.

That is the whole difference between AI doing the thinking and me thinking with it beside me. When I built those systems, I was the one deciding what good looked like. The tool moved faster than I could. But lean on it the other way, to skip the hard part rather than to reach further, and the same superpower slowly becomes a dependency.

What this means for the fear

There is a lot of fear about AI right now. Fear of being replaced, and fear that the skills a career was built on are about to stop mattering. I understand it, because I have felt versions of it myself.

But most of that fear is pointed at the wrong thing. The people I watch struggle with AI aren’t incapable; they are uncertain. They have been handed a powerful tool and almost no clarity about what it is for, and fear grows fast in a gap like that.

Because if someone who spent twenty years certain she wasn’t technical can build working systems in a month, then the stories the rest of us tell ourselves about what we can’t do are worth a hard second look.

The limit was rarely the skill. It was what I believed about the skill.