The LLM Job Paradox {N.B} Home Tags About The LLM Job Paradox<br>today 06/01/2026<br>· 5 min read<br>LLM · AI · Tech · Musings<br>Do you think AI can do your job?

By this I mean, do you think your entire job could be done by an AI. Can your entire team / field / department be removed without issue?

Generally, there’s two camps here:

AI can’t really do my job, but it can help me.

AI can take my job AND every other job, just give it time.

Does your management think AI can do your job?

Management here refers to people who are two or three layers above you, who’s day-to-day lives look completely different from yours. If you’re management, then you could ask yourself “do the people on the ground think I could be replaced with an AI?”.

The answer here is almost always “yes” and “they’re actively working on it.”

Do you think AI can take the jobs of your cross-disciplinary peers?

I am a software engineer, my peers would be designers, product managers, data scientists and so on. I could even split it into front-end vs backend-end engineers.

Most people believe that AI can do the job of their cross-disciplinary peers.

This is the paradox : the majority of people somehow believe tha AI can do the job of everyone else.

How is this possible? How can it be that the majority of the population believes that their work is unique/special but other work is not?

There’s another group of people that believe all jobs and work will be taken over by LLMs… that’s a different topic.

An old story

There’s an old story about Henry Ford that feels relevant here:

Ford, whose electrical engineers couldn’t solve some problems they were having with a gigantic generator, called Steinmetz in to the plant. Upon arriving, Steinmetz rejected all assistance and asked only for a notebook, pencil and cot. According to Scott, Steinmetz listened to the generator and scribbled computations on the notepad for two straight days and nights. On the second night, he asked for a ladder, climbed up the generator and made a chalk mark on its side. Then he told Ford’s skeptical engineers to remove a plate at the mark and replace sixteen windings from the field coil. They did, and the generator performed to perfection.

Henry Ford was thrilled until he got an invoice from General Electric in the amount of $10,000. Ford acknowledged Steinmetz’s success but balked at the figure. He asked for an itemized bill.

Steinmetz, Scott wrote, responded personally to Ford’s request with the following:

Making chalk mark on generator $1.

Knowing where to make mark $9,999.

Ford paid the bill.

— Charles Proteus Steinmetz, the Wizard of Schenectady

In my opinion, that story captures the essence of the paradox:

If you do not understand or truly appreciate the craft of someone, you end up reducing them to their output .

”Output” vs “Your job”

But, your job is not “making the chalk mark”. As a software engineer, we see extremely senior engineers spend days to produce a dozen lines of code while a junior engineer produces hundreds (even before LLMs). Or even, days to produce a negative output in lines of code.

Management likes to believe that all work can be reduced to a number1. But, we generally see their enthusiasm for reducing work to a number dwindle as we get higher up the chain. Sure, a CEO may be evaluated on their stock’s performance, but how much of that value is truly driven by their individual work?

Management has always been interested in measuring output, which necessarily reduces a worker to their volume of output. And LLMs are phenomenal at producing massive quantities of output.

The mimicry of work

Think back to a hard problem you solved, maybe some elegant code. You probably spent a fair bit of time on it, slept on the problem, before arriving at your solution. In your head, you iterated and played with ideas, testing them out before they ever manifested outside your mind.

The LLM is trained on your output, the final answer at the end. But it never saw the workings of your mind, it cannot mimic your mind because that data does not exist.

The LLM can draw chalk marks everywhere, but it will never understand “where to make the mark”.

Intellectual Empathy

Intellectual Empathy is the phrase I’d use to describe what’s missing here. Without it, we tend to reduce everyone to their output and ignore the actual intellectual labor involved.

LLMs are great tools for producing output when wielded by experts.

When we’re working in our area of expertise, it is immediately obvious the LLMs are not doing the intellectual labor needed to produce quality output. But, when we step outside of our areas, suddenly it feels “easy,” and that’s not because it actually is. It’s because we only understand the output.

As the allure of the LLM pulls you farther and farther out of your domain of expertise, it become a Dunning-Kruger amplification machine. Every idea (good or bad) gets spewed out, and without expertise in the field, there’s nothing in the loop to filter...