Hello! I'm your financial journalist and friend, an economist. The article we're going to discuss today actually hits a very sensitive point for many professionals and business managers.
In simple terms, the main argument of this article is that AI is making work outcomes cheaper, rendering the old method of assessing people's abilities based on their work ineffective.
Previously, if you presented a well-written piece of code or a detailed report, your boss would assume you understood the technology and the business. Now, AI can also produce such high-quality work, leaving your boss confused: Is this your achievement, or does it belong to the AI?
To help you fully understand this logic, I've broken down the article into five key points and explained them in plain language.
---
1. The old “portfolio” was a proof of ability; now it’s just a copy
Core logic: From “assessing ability by the outcome” to “the outcome no longer represents ability.”
For a long time, there was an unwritten rule in the workplace: Output equals ability.
- If an engineer delivered stable code, it was assumed they understood architecture and error handling.
- If a product manager wrote a logically sound plan, it was assumed they understood users and business constraints.
- If a consultant presented a well-structured PPT, it was assumed they had strong information-gathering and analytical skills.
Why did this logic work? Because the cost of producing these outcomes was high. To create good code or plans, one needed extensive learning, careful thinking, and trial and error. So, the outcome itself was like a trace of one’s ability. Companies couldn’t see what you were thinking, but they could see what you produced. If the outcome was good, it was likely that your ability was also good.
But AI has changed this balance.
Now, someone with average knowledge in a field can use AI to generate a well-structured report in just a few minutes; an inexperienced engineer can quickly complete tasks that used to take a lot of time with the help of AI tools.
The outcome is still the same, or even better and more professional. But the question arises: How much does this outcome really reveal about you?
In the past, the work was the original proof of your ability; now, it might just be a copy created by AI. With just a copy, it’s much harder for your boss to determine the true extent of your skills.
2. Stop worrying about “if you wrote it yourself”; ask “do you understand why you did it”
Core logic: Shift from “checking for cheating” to “testing comprehension.”
Many bosses, seeing that AI can do the work, react with panic and start implementing “anti-AI” measures, like asking, “Did you write this code yourself?” or “Can you do it without AI?”
The author believes this approach is a dead end.
It’s like asking a programmer to work without an IDE, search engine, or compiler to prove their programming skills. AI will eventually become as essential as electricity or the internet. Forcing employees to return to a world without AI is neither necessary nor realistic.
The real questions should be:
- Wrong question: “What percentage of this code did you write yourself?”
- Right question: “If the outcome deviates from the standard path, can you still find a solution?”
For example:
- Employee A: AI generated a design, but A can explain why that structure was chosen, identifies the weakest points, and knows how to modify it if the business needs change. If the model fails, A can fix it.
- Employee B: AI generated a design, and B simply says, “This is what the model suggested; I ran it, and there were no errors.”
Even if both A and B submit identical documents, A’s ability is much higher. The difference lies in who can handle the situation when the outcome goes wrong or changes. AI has reduced the cost of producing results, but it has also diminished the value of the outcome as a signal of ability. What companies need to judge is no longer just whether the task is completed, but whether the person truly understands, judges, and controls the outcome.
3. The most valuable abilities of the future lie beyond the delivered work
Core logic: From “executors” to “definers” and “gatekeepers.”
Since being fast and producing a lot is no longer rare, what abilities have become more valuable? The author lists four types of skills that were previously hidden by the outcomes but now need to be highlighted:
1. The ability to define the right questions (most valuable):
AI is good at answering “how to do it,” but not “what to do.” For example, if a customer wants an intelligent customer service system, AI can create one efficiently. But the real issue might be high costs or a disorganized knowledge base—these are often non-technical problems. The risk is that if the direction is wrong, AI can execute the wrong thing efficiently. Defining what needs to be done is more important than just executing it.
2. The ability to spot potential issues (suspicion):
Experienced people’s advantage isn’t in calculating faster than AI, but in knowing when to distrust the results. They know that the most dangerous part of a database migration is the rollback process, that attractive growth figures might be due to changed metrics, or that an AI-generated solution might not work in a production environment. These intuitions and experiences, once hidden in the work, now need to be explicitly demonstrated.
3. The ability to handle chaos (adaptability):
AI excels at handling well-defined problems, but the real world is full of uncertainty: sudden changes, missing data, inconsistent interfaces, and conflicting goals. The person who can reframe the problem when reality deviates from AI’s training data wins. The key question in the future is not “can you provide an answer,” but “can you remain efficient when the answer is not obvious.”
4. The ability to take responsibility for decisions (accountability):
AI can give advice, but it can’t take the blame for the organization’s choices. Companies need to know why a decision was made, what the basis for it was, and what risks were accepted. If something goes wrong, how can it be corrected next time? The future’s “advanced ability” is not about “how much content I can produce,” but “to what extent I can be responsible for the results generated by machines.”
4. The rules of recruitment and promotion need to be rewritten
Core logic: Shift from “focusing on output” to “focusing on quality and process.”
This shift will impact three core corporate mechanisms:
1. Recruitment: The traditional “take-home assignment” no longer works. Candidates used to spend a week on a project to demonstrate their skills, but now anyone can use AI to create a good result in a few hours. New approach: Interviews will be more like real-world scenarios, with fake requirements, unexpected situations, or hidden challenges. The goal is to see how you handle these, not to detect cheating, but to assess your problem-solving skills and ability to verify and challenge AI’s suggestions.
2. Promotion: “High output” no longer equals “high ability.” Writing a lot of code or documents used to be a clear indicator of merit. Now, with AI increasing output significantly, this becomes less relevant. New standard: Those who can reduce mistakes, identify risks early, handle ambiguous issues, and improve the overall quality of the team’s work are more likely to be promoted.
3. Performance evaluation: You can’t judge solely by output. If you only evaluate based on output, AI will make it easy for employees to achieve high scores by increasing the volume, but this rewards superficial success. New standard: The real challenge is to determine what should not be done, which outcomes are unacceptable, when to reject machine-generated suggestions, and when to stop. These subtle judgments determine the true quality of work.
5. A profound paradox: The easier the outcome, the harder it is to judge a person
Core logic: The concept of “ability” itself is changing.
In the past, ability was an internal attribute of an individual (I can code, I can analyze). In the future, ability will be about controlling systems—being able to manage increasingly powerful cognitive and execution systems (like AI) to produce reliable results from uncertain situations.
- **The key words are not “independence,” but “stability” and “reliability.”*
A person can rely heavily on AI and still be highly capable. The key is whether they understand what the system is doing, whether they can recognize when it goes beyond its limits, and whether they can take control if something goes wrong.
To summarize this paradox:
Getting high-quality results has become easier (thanks to AI), but judging why someone can achieve these results has become harder (because the process is becoming more opaque).
It’s like a software system that only returns a “200 OK.” You can’t see what’s happening inside. In the future, evaluating talent will require looking at logs, metrics, and tracing processes—to see how a person thinks, verifies, and corrects errors.
In one sentence: AI doesn’t make human abilities less important; it just shifts the focus from abilities that were once visible in the work (like writing or coding speed) to abilities that are now hidden (like judgment, problem definition, and responsibility). Companies must shift from evaluating work to evaluating the person, from asking for answers to asking for the ability to make informed decisions.