Advice is never neutral in an algorithm.
Mexico City, February 2026.
A viral genre has taken hold in career talk: ask an AI which degrees are not worth studying, then treat the answer like a cold economic verdict. The appeal is obvious, because it promises certainty in a world where tuition rises, jobs feel unstable, and prestige signals no longer guarantee mobility. Yet the deeper story is not which three majors get singled out, but why the question itself now feels reasonable to millions of people. When an AI ranks careers, it is not forecasting your life, it is compressing present labor anxieties into a neat list.
The first thing to understand is that these lists are rarely “AI wisdom,” they are a collage of prevailing narratives. Large language models echo the patterns they ingest: headlines about layoffs, salary surveys, employer complaints, and online cynicism about certain fields. That means the output tends to overreward what looks scalable, technical, and immediately monetizable, while underrating long horizon value and mixed skill pathways. In other words, the model is performing a social mirror, not a personal diagnosis.
Most of these recommendations are built on three simple priors: automation risk, wage compression, and market saturation. Automation risk is not only about robots doing tasks, it is about software unbundling professions into smaller pieces that can be outsourced or tool assisted. Wage compression follows when more people can perform the same output with fewer hours, or when employers can replace entry level labor with tools. Saturation then appears when universities produce graduates faster than local economies produce roles that reward that credential.
The problem is that these priors are partially true and still misleading, because they treat “degree” as destiny. In reality, outcomes depend on specialization, geography, networks, internships, portfolio evidence, and the ability to hybridize skills across domains. A generic label like “communication,” “business,” or “law” hides enormous variance between low signal credentials and high signal training inside the same discipline. AI lists collapse that variance, then present the collapse as clarity.
There is also an embedded bias toward visible labor market signals, which are usually short term and unevenly measured. Fields with clear job titles and standardized salary ladders look safer to an algorithm because they are easier to quantify. Fields that produce diffuse outcomes, like entrepreneurship, public service, arts, or research, look riskier because the payoff is nonlinear and often delayed. The AI ends up rewarding what is legible to metrics, not what is valuable to a society across decades.
Another hidden issue is that “do not study this” often means “do not study this in the cheapest, most generic, least differentiated way.” Many degrees become risky when they are pursued without a strategic edge: no internships, no technical complement, no writing samples, no projects, and no signal of competence beyond the diploma. The degree is not the asset, the asset is the capability stack that the degree can anchor. When AI discourages entire fields, it often confuses a weak pathway with the discipline itself.
If you want to use an AI answer intelligently, treat it like an input to due diligence, not like a verdict. Ask what assumptions are driving the ranking, and whether those assumptions apply to your context, your region, and your intended niche. Then pressure test the degree by mapping it to real roles, required skills, and measurable outputs you can build while studying. A credible plan should show how you will graduate with proof of ability, not only with credits.
A more realistic framework is to evaluate programs through four lenses: adaptability, portfolio, scarcity, and resilience. Adaptability asks whether the training helps you learn new tools and switch domains as industries shift. Portfolio asks whether the program produces artifacts you can show, like projects, case work, code, research, or documented outcomes. Scarcity asks what you will know or do that is uncommon among graduates in your market, which is where bargaining power lives. Resilience asks whether the path still makes sense if the first job is not ideal, because early career turbulence is now normal rather than exceptional.
The final point is psychological and strategic: outsourcing life decisions to a machine can feel calming, but it can also shrink ambition into risk avoidance. Many of the most meaningful careers are not the “safest” on paper at age eighteen, and many high paying tracks become traps if they do not match temperament and values. AI can help you see structural pressures, but it cannot tell you what you can endure, what you can love, or what you can build over time. The right use of these lists is not to obey them, but to convert them into sharper questions that force you to design your education like a strategy, not like a tradition.
Narrative is power too.
La narrativa también es poder.