Selectel: AI isn't taking jobs, but it is making it much harder to enter the profession

AI does not abolish work; it changes the rules of access to it. That is the conclusion Selectel reaches in its analysis of how artificial intelligence is reshaping the labor market: the number of jobs is not shrinking, but the way into the profession is narrowing, requirements are rising, and the price of AI-related skills is climbing sharply. Instead of a scenario in which machines displace people en masse, a different picture is emerging: people are being replaced by those who have already learned to work alongside models and turn AI into practical results faster.

Over the past few years, a whole layer of new roles has appeared around AI that either did not exist until recently or were niche. At the center of this ecosystem is the LLM/AI Engineer, who does not train a model from scratch but assembles a working system from APIs, RAG, tools and pipelines. Next to them stands the updated MLOps engineer: the job is no longer just to roll out models, but to keep latency, reliability and inference cost under control.

Higher still in scarcity and pay is the AI Infrastructure Engineer, a specialist in GPUs, distributed computing, quantization, caching and other things that directly affect the economics of an AI product. The AI Product Manager and the AI Interaction Designer round out this layer: the former balances answer quality, request cost and user experience, while the latter designs the logic of the dialogue, the system's behavior and how it should fail.

A separate group of roles concerns data, quality and safety. Prompt Engineer, once a fashionable standalone profession, is gradually turning into a set of required skills for engineers and product teams. Human-in-the-loop specialists and AI Trainers label data, correct model answers and in effect create the hidden human layer beneath "automatic intelligence." The Synthetic Data Engineer goes further and generates training sets for rare or poorly represented scenarios. AI Auditors and AI Risk Specialists check systems for bias, legal risks and regulatory compliance. In other words, the market is not simply adapting old professions to new tools; it is building a full production infrastructure around AI.

But as new roles appear, the hiring principle itself is changing. According to data Selectel cites from research, up to 66% of companies are cutting hiring of specialists who would need long retraining, and about 90% of employers report the disappearance or radical transformation of basic roles. This hits juniors especially hard: boilerplate code, initial analytics, report preparation and other routine work are increasingly automated.

Companies are less willing to take someone on for future growth and prefer candidates who already use AI tools in their workflows. Against this backdrop the market is polarizing: the top segment, covering engineering, infrastructure and AI product management, is getting more expensive; the bottom is being crushed by automation; and mid-level people are stuck in the middle, with some of their tasks disappearing and the rest demanding a nearly senior level. At the same time, the premium for AI competencies is growing: such skills can, by some estimates, add up to 15% to a salary.

Layoffs have not gone anywhere, but they need to be read correctly. Selectel cites an estimate that in 2025 the tech sector globally lost about 246 thousand employees, of whom roughly 55 thousand layoffs were linked to AI to some degree. In the first months of 2026 the trend continued: about 40 thousand more job cuts, with the average rate rising from 674 to 926 people per day.

Still, AI is rarely named as the sole cause of layoffs. Companies usually talk about restructuring, efficiency gains and reallocating resources toward AI initiatives. It also matters that AI's direct contribution to layoffs remains limited for now: in early 2026 only about 7% of cases in the US were tied to it, and chief financial officers on average expected headcount to fall by only about 0.

4% over the year. This suggests AI more often acts as an accelerator of optimization already under way, not as the sole source of crisis.

The main conclusion is not that there is less work, but that the labor market has become markedly more selective. The winners are specialists who understand the limits of models, can calculate inference economics, build reliable pipelines and take on harder tasks than before. What loses out is not the profession itself but the old way into it: through simple tasks, a long ramp-up and gradual learning on the job. That is why the main competitor today is not AI as such, but a person who has already built it into their daily work and learned to get measurable results from it.