Artificial intelligence multiplies the capabilities of experienced professionals, but it may be weakening the path that allows young people to acquire that same experience.
In the era of AI and agents, few professional groups seem as exposed to this transformation as recent graduates, especially from fields where this revolution has hit hardest, such as computer science and law, but it will likely extend soon to other service-industry-related careers.
In my professional experience, the tasks we call "agentic" are better coordinated and supervised by experienced or senior professionals. The reason is that they are able to align agents more effectively with the objectives of the problem to be solved, give better instructions, and make more accurate critiques of the results obtained. Thus, both the agent and the senior mutually enhance each other; corrections allow for refining instructions and future processes, while the senior absorbs the acquired knowledge in a close symbiotic relationship.
Until recently, companies delegated non-critical but necessary tasks to novice workers, such as repetitive labor or information gathering and sorting. Today, many of these minor tasks are performed by AI in seconds, drastically reducing the opportunities newcomers had to demonstrate their professional skills, gain experience, and improve their standing among colleagues and managers.
A junior professional, while bringing fresh knowledge and energy, lacks the experience and judgment provided by professional practice. Additionally, what is taught in universities and institutes is not always directly applicable in the real industry. When a junior operates with AI/agents, they often do not question them sufficiently, and when they do, they lack the necessary criteria. While superficially results might appear equivalent between juniors and seniors using AI, recent studies like one conducted by Google Research show that this shortcut can improve the quality of a beginner's work today, without necessarily translating into the development of the mental "muscles" and critical judgment that will make them an expert tomorrow.
Furthermore, some projects do not end with the deliverable; their real outcome is measured in the medium/long term during production operations. This is the case with cloud software platforms or SaaS, where long-term customer retention is even more important than the sale itself. In these projects, seniors excel because they not only guide the agents they program but also tend to simplify and standardize the architecture to minimize future operational costs, including the probability of errors and service unavailability. This is an exercise that only the confidence of expertise can provide, as it often involves contradicting the AI. A junior usually accepts and trusts what the agent does more easily, giving vague (or overly specific) instructions and auto-accepting paths and branches that unnecessarily complicate the software, increasing operation, maintenance, and future change costs.
This is undoubtedly a tremendous opportunity for experienced professionals: not only are their capabilities multiplied, but their job security will likely strengthen in the coming years, considering that responsibility for work done with AI will continue to fall on people and organizations, from a company's legal representative to a department head. And there is a simple explanation: you cannot legally punish or sanction an AI for poor work: there will always be a human taking responsibility (for now).
But what will happen to young professionals? A generation facing growing difficulties entering the labor market and, in times of misinformation, also exposed to manipulation, will hardly be able to escape alone from a crisis that could worsen with the expansion of AI into more industries and applications. We must keep in mind that frustration, forced free time, inexperience, lack of real responsibilities, and absence of life perspective can become, under certain circumstances, the perfect fuel for riots and social unrest. The question remains open.
