Will AI Replace You (Yes, You) in the Near Future?

A-a-a-a!!!111 We're all going to be fired! - one part of the internet panics. The second part is secretly afraid, but hopes for the best. The third part goes to fix the pipes or wiring at their neighbors' houses, and sighs - I wish AI would replace me so I could finally spend time with my grandchildren. And the fourth part thinks that if they surround themselves with LLMs and agents and start flooding social media feeds with posts on this topic, they'll get away with it.

This article is an attempt to analyze the prospects of replacing humans with AI depending on the type of occupation.

So, professions are divided according to interaction:

• Human ↔ human: baristas, waiters, doctors, politicians, managers.
• Human ↔ tool: turners, fitters, plumbers, builders.
• Human ↔ computer: programmers, designers, copywriters.

Some professions combine these interactions. For example, a doctor may look at CT scans of patients or read blood test results on a computer and make a diagnosis based on them, or communicate with the patient, reassuring them that a slight runny nose is unlikely to kill them in the near future.

So which of them will AI replace? Let's consider each option:

Human ↔ Computer

Obviously, those who interact exclusively with computers—designers, programmers, testers—are the first to be affected. If the job is related to computers and nothing else, then such an employee is, as offensive as it may sound, an interface between the task and the computer on which they perform that task. Moreover, the task is also most often assigned via a computer and is formal. And the less such an employee interacts with real people, the more likely they are to be replaced by a neural network.

This is because there is nothing to prevent their manager from setting the same tasks via LLM and monitoring their completion. The only question is the sophistication of the model and the accuracy/detail of the task setting.

If, on the other hand, the employee focuses more on human communication, for example, by understanding the domain, clarifying the task, highlighting potential problem areas or edge cases, clarifying concerns, and taking responsibility for a certain function (i.e., moving to human ↔ human interaction), then the chances of them being fired decrease.

Why:

• The employee's visibility to their manager increases
• Their expertise in the subject area grows
• The manager sees such an employee as one of the key members of their team, someone they can “rely on” in a crisis

Of course, no one is immune to the possibility that their entire department could be eliminated. But if the layoffs are selective, the chances of staying will be much higher than for their sociophobic colleagues. Even if the entire team is laid off, the connections built up during their time at the company will help them find a new job much more quickly and easily.

Conversely, if an employee does not communicate with anyone, remains silent at meetings, and sits with their camera turned off, then even if they perform their tasks well, such an employee is more likely to be a candidate for replacement. As sad as it may sound. Because no matter how good a person and employee they are, business puts money first. And obviously, paying an LLM provider $200 a month is simply more profitable than paying a developer at least 10 times more.

Human ↔ Tool

You have to admit, it's hard to imagine a situation where AI could replace, say, a bricklayer or a furniture assembler. No, theoretically, of course, it's possible to imagine and even implement such a thing. But it just wouldn't be profitable. A robot with the necessary skills would cont way more to operate than a human. Not to mention the cost of development. In addition, humans are much better at adapting to non-standard situations. For example, if something doesn't fit properly for a furniture assembler, they can quickly modify the necessary part right at the customer's place, and everyone will be happy. Or, a plumber may find during installation that a certain fitting needs to be replaced. For a robot to be able to do that... Well, I don't know, we'll have to make progress in that area in at least a decade, but I suspect it will take much longer. So, AI does not pose a threat to replacing skilled workers yet. For now.

Human ↔ Human

Finally, we come to the most interesting part. Thousands of books have been written about human interaction, and it would be nearly impossible to summarize them all in a single paragraph. It would be like trying to push a log through the eye of a needle. But let's give it a try.

No AI can replace human communication. Some may argue that AI can replace a psychologist with whom we communicate, for example, via video, and that would be perfectly fair. But here we return to the interaction between humans and computers. Intonation, a joke that accidentally pops up out of context, a handshake, a shared lunch. I am talking about nonverbal communication. Models cannot reproduce this yet.

And in the end, who fires employees after the implementation of AI? The same guys who play golf with each other and decide strategic issues over a glass of cognac — whether it is profitable to replace a developer or sales manager with AI.

Summary

In summary, the more live communication is involved in your work, the less likely you are to be replaced by AI or an agent. And blue-collar jobs are also out of the danger zone for now.

Of course, with the development of AI and robotics, just about anyone could be affected. But in some cases, it may simply not be profitable, and in others, it could create social problems, such as mass unemployment, and then government regulators will step in.

Of course, prediction is a thankless task, and no one knows what will happen even in five years. And black swans have been arriving with enviable regularity lately. Although, maybe we've just become better informed.

Comments

[Bhavin Sheth]

Good breakdown. From my own experience building tools, AI replaces small repetitive tasks first — not the person. The real value is in understanding the problem, making decisions, and taking responsibility. I use AI daily, but it still needs direction. People who adapt and learn how to work with AI will be fine.

[Maxim Thomas]

Thanks, I agree. Even if AI learns to direct itself, all AI-generated work should still be supervised.

[John Samuel]

Nice framing with the three interaction types. I’d add that AI doesn’t just threaten ‘human ↔ computer’ roles, it also reshapes them: the people who learn to orchestrate tools, talk to stakeholders, and own outcomes become force multipliers rather than ‘interfaces.’