Garbage In, Genius Out? The Truth About AI and Expertise
There is a version of research that looks like this: you collect data, you run the numbers, you present the output. Clean. Mechanical. Repeatable. And increasingly, with the right AI tools, almost anyone can do it.
That version has always bothered me.
Not because it’s wrong, but because it’s incomplete. Because the researchers I find most interesting, the ones whose work actually moves things, are doing something harder and less visible than that. They’re making judgment calls. They’re deciding which variables matter and which don’t. They’re reading a situation and knowing, before the model confirms it, that something important is hiding in the noise.
That’s the conversation I wanted to have. So for the Alumni Edition of Hult Research Lab, I brought in two people who are living proof that research done well is one of the most demanding intellectual disciplines there is.
Samuel Harris, MBA, MSc, is a quantitative analyst working at the intersection of mathematics and financial markets. His job, in his own words, is to apply mathematical, statistical and machine learning research to financial markets. The kind of work most people encounter only in movies, and usually don’t fully understand even then.
Thamsanqa Ndlovu (Thami) is a senior market researcher and founder of Datadvise, a consultancy he built independently over the last six years. His background is in law, and his work spans solar energy deals in Southern Africa, mental health policy, competitive intelligence, and economic development advisory. On any given Tuesday, he might touch all four.
Two different disciplines. Two very different career paths. And yet when I sat down with both of them, the same three ideas kept surfacing, each from a different angle, each landing in the same place.
The first is that narrative is not the packaging around research. It is the research.
Sam made this point from the quantitative side. AI tools, he argued, are only as useful as the expertise of the person using them. If you don’t understand the underlying story of what you’re analyzing, you can’t ask the right questions, and you won’t catch the wrong answers. Thami arrived at the same conclusion differently: a consulting deliverable that doesn’t build toward a coherent story is just paper. The numbers have to mean something. They have to point somewhere. This isn’t a philosophical preference. It’s the practical difference between work that changes decisions and work that gets filed away.
The second is that intellectual restraint is a skill, and most people underestimate it.
Sam’s instinct when facing a new problem is to ask how little complexity the problem actually requires. His current project, solved with a closed-form mathematical solution, could have been built with machine learning. He chose not to, because the simpler solution was equally accurate and significantly more efficient. The fancier option wasn’t better. It just felt better. Thami operates with a similar discipline. His seventy percent rule is a deliberate defense against the trap researchers know well, the belief that one more data point will finally make the decision obvious. It won’t. At some point, you have to trust your directional clarity and move.
The third is what AI actually does to expertise, which is less than most people think, and more specific than most people admit.
Both of them pushed back against the idea that AI meaningfully democratizes high-quality research. What they described instead is something more precise. AI raises the floor. It compresses the gap between nothing and average. But the gap between average and genuinely good, that still requires accumulated judgment, domain knowledge, and the ability to recognize when a confident-sounding output is quietly, completely wrong. If you don’t have the expertise to notice, you’ll never know it happened.
What connects these three ideas is a conviction I share, that good research is not primarily a technical problem. It is a thinking problem. The tools matter, but they are downstream of something harder to build: the ability to ask the right question, hold a process with integrity, and know when you have enough to act.
That is what this podcast is about. Not AI as a trend. Research as a practice. Thinking as a discipline worth taking seriously.
This Alumni Edition is live. Go listen!
Check it out on YouTube!
Connect on LinkedIn for more consistent updates!
— Max Getuba
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit theimpactfulcapitalist.substack.com [https://theimpactfulcapitalist.substack.com?utm_medium=podcast&utm_campaign=CTA_1]
