AI Brought Back the Joy of Making

Conversations with a bio researcher, a finance data engineer and a big tech designer

The story of AI changing how developers work is old news by now. What I wanted to know was how AI is being used by people in completely different fields: someone editing genes in a lab, someone handling financial data on Wall Street, someone designing interfaces in a big tech R&D group. And not in Silicon Valley, but in New York and Boston.

The first thing all three had in common: none of them could ever go back to working the way they did before AI. They were all already using it heavily. For the bio researcher, AI was a tool that changed the speed of research. For the finance engineer, it was a force reshaping how the team is organized. For the designer, it brought back a joy of making she had lost. Same technology, a different meaning for each person.

My earlier posts were mostly about how engineers are taking up AI. This time I'm writing about interviews with non-developers in three different industries.

1. AI that edits genes

In Boston I met an AI researcher who works at a life sciences institute run jointly by MIT and Harvard. They build AI tools for bio research. They've published papers on CRISPR, the gene editing technology, and they spend every day where life science and artificial intelligence meet.

When I asked about AI's potential in biology, the answer was firm: "Only about 30% of life science is understood so far. Even that's an estimate. Either way, it's still largely unknown territory." With software, I've sometimes wondered how much is really left to do now that AI exists. Talking with this researcher, I came away thinking that bio and space may be the fields where R&D finally takes off because of AI. They said there's still so much for humans to explore that there's a huge amount of work ahead.

They also told me they'll soon be moving to a new bio AI company. It's a lab founded by the professors who helped map the human genome, and it's going all in on AI, building its own AI models, with the goal of curing every intractable disease humanity still faces.

2. An AI betting bot as a side project

What I found even more interesting was their side project. They were using Claude Code to build a betting bot for Polymarket, a prediction market platform. Bio AI research by day, coding a financial prediction bot after work. There's a reason this combination is now possible.

"It feels like I code about three times faster." They're not a pure developer by background. They came up as a researcher, someone with domain knowledge and analytical skills, but coding was never the main job. Now they build working bots on their own. It's a textbook case of AI lowering the barrier to coding. They already knew how to read data and form hypotheses, and Claude Code served as the bridge that turned that ability into actual code.

3. A data engineer on Wall Street

In Manhattan I met a data engineering team lead at a global financial data company. Ten years into their career. They started out at an ad agency, joined this company as a senior data analyst, then moved into data engineering and worked their way up to team lead.

Finance is a conservative industry. They only started using Claude Code recently, because strict security rules slowed the adoption of outside AI tools. Once they started, though, things changed fast. They said life as a manager in particular is different now. The time it takes to summarize the team's work, write documentation and handle communication has dropped noticeably.

Most financial companies these days are also building their own AI models. That data represents decades of accumulated know-how, and they don't want to depend on outside tools. (It's not like they're short on money.) Financial data is also complex in structure and held to a high bar for accuracy, so general-purpose models only go so far. That's why most of the big financial firms are going the route of developing their own models.

4. The outsourced grunt-work teams are disappearing

The thing that struck me most in this conversation was how many outsourced teams there were doing simple manual work, in Southeast Asia, India and China. Financial data turns out to be more unstructured, and shared in less friendly formats, than you'd expect. In those cases, people first had to read the data by hand, pull it out and transfer it over.

"Disappearing" was the right word. They didn't vanish all at once. They were shrinking gradually. As AI took on more of the work, those teams got less volume, and the contracts naturally got smaller. Finance still has a strong Excel-based culture. But who is doing the work inside those spreadsheets is changing. AI has started taking the place of the people who used to fill in the cells.

From the manager's side, this actually made things easier, since the overhead of communicating with those teams, checking quality and coordinating schedules went down. The teams themselves probably saw it coming.

5. Now people use ChatGPT to calculate the tip

The funniest part of this conversation was about AI in everyday life. New Yorkers now use ChatGPT instead of a calculator to figure out the tip at restaurants. That's how far AI has gone beyond a work tool and into every part of life.

Another example: they recently ran their own apartment lease through Claude. A New York apartment lease runs dozens of pages and is full of legal jargon. In the past, you'd either just trust the other side and sign blind, or pay a lawyer a lot of money to review it. Now Claude summarizes the key clauses and points out what to watch for. It doesn't fully replace a lawyer, but in their words, "at least you know what to ask."

They were specific about where they use AI most as a manager: summaries after team meetings, project documentation, and briefings for reporting up the chain. As a manager they hardly write code themselves anymore, but AI saves them a lot of the time spent on communication work like this.

6. A designer who spends $500 a month of her own money on AI

She's a product designer with ten years of experience at an R&D organization inside a US big tech company. Unfortunately, her company doesn't officially provide AI tools. So she spends $500 a month of her own money on Claude tokens.

When I asked why she goes that far, her answer surprised me: "I started feeling the joy of making things again." After more than ten years as a designer, at some point the work had become routine. Draw the wireframes, get feedback, revise, get feedback again. Somewhere in that loop, the joy of making slowly faded. AI brought it back. When she refined her work together with AI, she could take it to a level that was hard to reach on her own. Along the way she found herself working late into the night. That had never happened before.

$500 a month is not a small amount. But she told me, "With this money, I got something back." The question for her was how she wanted to feel during the remaining ten years of her career. For her, AI wasn't an efficiency tool. It was something that protected her identity as a creator.

7. Why AI-made UI feels off

What I was most curious about was the difference between designs made by humans and designs made by AI. Her first point: "UI made by AI doesn't have a consistent design system." Each individual screen looks fine. The buttons are pretty and the layout is clean. But put several screens side by side and the color tones are subtly different, the spacing rules aren't consistent, and there's no visual language running through the whole thing.

A regular user might not notice. But building that detailed consistency across a whole service is a designer's core job, she said, and it's still hard for AI to do. It's the instinct that lets ten years of experience spot in thirty seconds that "something here is off."

She described LinkedIn as "a little stream of anxiety." The US and Korea are pretty similar there. Every day you open the feed and there's new AI news, someone's review of the latest tool, and so on, and you feel a faint pressure watching it all. The FOMO of wondering if you're the only one missing out. In that situation, she said, the anxiety goes away once you actually build something with your own hands. As you build, you develop your own know-how and get insights that go beyond the news.

8. The lines between roles are breaking down

Put the three stories side by side and one shared pattern shows up: the lines between roles are blurring.

The bio researcher builds a betting bot. The engineering team lead uses AI to write documents and analyze contracts. The designer has started learning to code. Everyone's job title is the same, but the range of what they actually do keeps growing.

All three of them also said the same thing: there are things AI still can't do. The bio researcher said, "AI can't decide what to do next." A human would naturally finish a task and move on to try something else. Analyzing experiment results and deciding the next direction, forming hypotheses, judging which questions matter. That's still the human's job. The finance data engineer said something similar. AI is good at extracting and organizing data, but deciding "what decision should we make with this data" was up to people. Same for the designer. AI can generate screens, but setting the direction for the whole product is a person's job.

The tools have become powerful, and the range of what people can do with them has grown. But the ability to set direction, to know what should be built, was becoming more and more of the deciding difference.

The designer shared a portfolio she'd built purely through vibe coding, and wrote back that AI was letting her do things she never could have tried before. If anything, AI helps designers with more experience.

After meeting the three of them, one thought stuck with me. In the end, all three were recovering the joy of making through AI, each in their own way. For the bio researcher, it was the joy of turning theories that lived only in their head into code that actually runs. For the finance engineer, it was the joy of redesigning work that dozens of people used to do by hand. For the designer, it was the joy of breaking out of a ten-year routine and wanting to stay up all night polishing something again.

Whether AI is a threat or an opportunity is already a stale question. For these three people, AI was neither. It gave back something they hadn't even realized they'd lost: their sense of themselves as people who make things. The people who get that back will keep moving forward.

Originally written in Korean on Brunch, April 13, 2026.