Article
Can AI Replace Human Judgement?
AI can now produce the graphic for an advert. It writes the newsletter. It prepares the product page. The person who used to do this work is probably still sitting in the office. The question is what exactly their job is, and what it will be.
For a long time the story was that AI would take over the routine and leave people the important part. It does not have to be that way. AI can work out faster than a person where a company has the most potential to move revenue, acquisition, retention. And it can carry out the steps it proposes itself.
When AI makes both the decision and does the work, what is left for the person?
The obvious answer is judgement.
What Judgement Looks Like in Practice
It is the ability to look at an output and know whether it is good — and to know it in the context of a particular company, not in the abstract.
Take the work of an advertising specialist. AI generates ten graphics for them. If that person has no judgement about what a good advert graphic looks like in the context of that specific company — its customer, its brand voice, its market — then they cannot be the one whose judgement decides. They click on the one they like. That is not the judgement I mean. That is subjective taste with nothing behind it.
The same holds for the newsletter. For the product page. For anything that can be delegated to AI. The value of the person has moved from producing the thing to judging it. And only someone with real expertise, anchored in the context of the company, can judge it.
Why Most People Lack Judgement
Here is the uncomfortable part. Most people working in a given field today have worse judgement than AI. Not better. Some of them know it, and so they delegate the judgement too — they let AI decide what is good as well, because they cannot judge it more reliably themselves.
But when a person cannot judge quality properly, they cannot meaningfully delegate the work to AI either. To delegate is to stay responsible and able to check. Someone who cannot check only hands the work over and hopes — and their position loses its meaning.
In many companies you can see it clearly. There is someone who only picks from what AI generates, by feel, and changes nothing about the result. And there is someone who is critical of the result, and whose prompting actually gets the output corrected. That is the dividing line.
The minority on the other side of that line are different. Their judgement — anchored in the context of the company — is often far better than anything AI produces. And getting AI to their level is exactly the hard part.
Can AI Be Taught Judgement?
This is the crucial question. My answer: yes, but not in the way most people imagine, and never with a single prompt.
Join the Library
Full access to my thoughts, personal stories, findings, and what I learn from the people I meet.
Join the Library — €29.99 per yearSummary
Common questions on this article's topic
Can AI replace human judgement?
Can AI be taught a company's judgement?
Why is internet training data not enough for judgement?
Whose job does AI replace first?
How do I know if my judgement is replaceable?
If you have any thoughts, questions, or feedback, feel free to drop me a message at mail@richardgolian.com.
Related articles
Yesterday I could not tear myself away from the computer. When I lifted my head, it was half past eight in the evening. I had been sitting alone upstairs for about three hours.
Will AI take my job? A certified Google trainer told me in June 2024 that my profession would cease to exist. Twenty-two months later, my job title has not changed — but ninety percent of what I do during the day is different. I have delegated more of my thinking to AI agents than I thought possible. I am not afraid. This is why, and what it means for anyone asking the same question.
In April, in the first part of this series, I wrote about an AI prediction system I had started building on my own machine. At the time the software was a few hours old and the prediction record was empty. The record since then has shown one thing — the system does not yet understand the market it is being asked to forecast. It can pull macro context, book value, earnings. But it cannot put those together into something that helps it understand the price.
More articles
Prague, 13 May 2026. On my way to work I started thinking about something that stayed with me for days. If most routine work on a computer disappears in the next ten years, and a large share of repetitive manual work disappears with it, what happens to the flow of money? Who pays whom for what? Which economic layers will exist, how large will they be, and what relationships will run between them? This is the six-layer map I sketched as an answer.
I am building an AI system to predict the S&P 500. It runs on my own machine, uses free public data — yfinance, FRED, the Shiller dataset — and grades every forecast against reality. This series documents the build itself: the decisions, the methodology, the mistakes. What I will eventually share from the running system is a separate question, and an honest one.
One hour. Fifty-five minutes. That is how long it took to build what a Czech software firm had quoted at over €50,000. I built it with Claude Code. Not a prototype. Not a proof of concept. A working tool — the one the company actually needed. By the evening of the same day, it was running on staging. This is not about Claude Code. It is about what Claude Code exposes.
I have conducted roughly one hundred and fifty practical interviews over the past four years. Fifty for data specialist roles. A hundred for advertising and performance marketing specialists. Almost every one of them involved sitting down with a candidate over a practical task — something close to a real problem we actually need to solve at the company. Not theory. Not trivia. Applied problem-solving. Over time, I started noticing a pattern.
Before you can teach AI to understand anything, you need to see what it is hiding from you.
The moment other people needed access to it, the problem changed completely. It was no longer about whether the agent could learn. It was about who gets to teach it.
I wanted to build an agent that doesn't just assist. One that acts.
Four days in Catalonia. No computer, no AI, almost no social media. I bought this notebook so that I could write down what I would think about, and what I would come across and learn on the trip.