Skip to content
Artificial intelligence · Atlassian · OpenAIEpisode 144 · 4 August 2026 · 25:20

AI Is Changing Professions Faster Than Companies Can Redesign Work. What Should People Do?

Central question

How will the next experienced professional emerge if AI takes over the entry-level tasks on which lawyers, analysts, programmers, and designers used to learn?

What you take away

Understand why access to AI does not replace the path to competence and how to redesign the first steps of a profession so that people learn to reason, verify, and take responsibility for outcomes.

Main threads

What to watch for

1Preserve independent tasks for beginners in which they must formulate an approach before turning to AI.
2Evaluate not only the finished output but also the explanation of the decision, source verification, and the ability to spot a model error.
3Give junior specialists real responsibility with mandatory review, not merely the role of prompt operator.
4Watch whether AI-enabled role expansion is becoming simple workload intensification without growth in skills or authority.
Signals to track afterwards
The decline of junior openings and changes in the set of entry-level tasks within professions.
New training programs that combine AI use with independent analysis and verification.
Rising expectations for one employee without corresponding changes in title, pay, or authority.
Which companies continue developing professionals and which rely only on a senior-plus-AI model.
Most useful for
students and early-career professionalsteam leadersHR and learning professionalseducators and education productspeople planning a career change

Key takeaways

00:00AI Takes Over the Tasks People Used to Learn On

Entry-level work often consisted of drafts, research, basic analysis, and repetitive operations. These are the tasks the model automates first.

06:05A Fast Result Can Hide a Lack of Understanding

A learner receives a finished text or solution before developing their own reasoning. Output quality rises faster than the person’s competence.

07:45Access to a Tool Does Not Create Knowledge

ChatGPT expands capability, but the person still has to understand the field, detect errors, and own the outcome. Using the system does not replace the path to experience.

10:38The Labor Market Changes Entry Before the Profession Itself

The job title may remain, while the number of junior positions and the first set of tasks already change. The transition begins at the bottom of the career ladder.

13:41Role Expansion Can Mean Work Intensification

When one person with AI performs more functions, this does not always mean greater freedom. A company may simply raise expectations without allowing time to learn or granting new authority.

15:12Without a First Step, the Next Senior Never Appears

A senior-plus-AI arrangement may be efficient today, but if beginners receive no real work, companies will have no one ready to replace experienced people in a few years.

17:37The First Step Must Be Redesigned

A beginner does not need an AI ban, but tasks in which they first form a solution, then use the model, and finally examine the differences.

21:39AI Must Become Part of a Profession, Not a Substitute for It

Education must connect domain knowledge with model use. Otherwise, a person learns to obtain answers without understanding when those answers cannot be trusted.

What this episode is about

If the first tasks of a lawyer, analyst, programmer, or designer are now handled by an experienced worker using AI, where will the next experienced specialist come from? Anna Volchek explains why access to ChatGPT does not replace the process of building competence.

AI Is Taking the Tasks People Used to Learn On
The first years in a profession rarely involve the most difficult decisions. A junior employee collects data, prepares drafts, checks documents, writes simple code, and observes how an experienced colleague works. Those are precisely the tasks AI performs most quickly. For a company, this looks like a productivity gain: one strong employee using ChatGPT can handle work that previously required several people. For education and the labor market, however, it creates another question—where the first step into the profession now takes place.

Access to AI Does Not Create Knowledge Automatically
A person can receive a good text, analysis, or program without understanding why the result is correct. That expands capability while hiding gaps. If a junior specialist cannot independently verify the model’s work, they may look productive without actually building competence. The use of AI in education therefore cannot be reduced to permission or prohibition. The real question is what the person still needs to learn when the system already performs part of the exercise.

Role Expansion Can Become Work Intensification
When one employee gains the tools of several professions, their role genuinely expands. But a company can use the same effect differently—not to give the person more autonomy, but simply to increase the workload. In that case AI does not free time; it compresses more work into the same day. This conflict matters for the labor market because higher individual productivity does not automatically improve work quality, learning, or career opportunity.

The New First Step Must Combine AI with Independent Verification
The next experienced specialist will not emerge simply by learning to formulate a good request. They still need to distinguish good work from bad, understand the domain, make decisions, and take responsibility for them. AI will therefore have to be integrated into every profession as a working tool, while preserving tasks in which the person reasons independently and can demonstrate the quality of the work. Otherwise, the market will produce many people who generate output quickly and too few people who can evaluate it.

The central risk to professions is not only job loss but the disappearance of the path from novice to expert. AI must be integrated in a way that accelerates practice without taking away the development of a person’s own judgment.

Episode transcript

The episode is in Russian; below is an English reading guide to the transcript (the full EN transcript is a machine translation). Voice matching applied to 35 segments: 35 identified, 0 mixed, 0 marked with ✓, and 0 unresolved.

Loading…