90% of Americans fear AI, but universities are ramping up enrollment in AI degrees

By Zendoric · September 26, 2026.

The starting figure looks like a contradiction: nearly 9 out of 10 Americans say they distrust the advance of artificial intelligence, according to a recent CNN poll reported by Spectrum News. And yet, on the country's university campuses the opposite is happening. According to the Center for Inclusive Computing, 74 US universities already offer a specific major in AI and 94 have a specialization or minor, and a good share of those programs have been created in the past year alone.

The case the report uses to illustrate this is MIT (Massachusetts Institute of Technology). Its "AI and Decision-making" program, launched in 2022, has grown from a few dozen students to more than 300 and is now the school's second most sought-after major. Julia Schneider, one of the first students to declare that major and now a graduate student, says ethics and safety are not an appendix to the curriculum: there are entire semester-long courses devoted solely to that, and even her informal conversations with classmates —she says— are 80% about AI safety. Tracy Camp, executive director of the Computing Research Association, confirms the trend is widespread: there is "tremendous popularity" in AI majors, minors and specializations across the country.

What is striking is that the momentum comes not despite the industry's warnings, but partly because of them. The report itself recalls that, after the resignation of an Anthropic researcher who warned of the technology's risk to humanity, several industry leaders called for slowing development and lawmakers pushed new oversight rules. Schneider reads it the other way around: "now is precisely the time to study more AI, not less," because what is needed are people focused on understanding why these models do what they do.

That reading is not naive; it is consistent with how talent markets behave in the face of a risk perceived as real: those who best understand a technological danger usually want to work on governing it, not fleeing it. It happened with cybersecurity after the big breaches of the past decade, and it is happening now with AI. It also fits something we were already observing in our analysis of AI's impact by sector: in technology, the winners are those working on architecture, security governance and the integration of these systems, while purely routine work loses ground. That the generation entering university today is choosing en masse to train in exactly that is not an anomaly set against society's fear, it is the flip side of the same coin: fear creates demand for professionals who know how to handle it.

That said, it is worth keeping an eye on the short-term risk this rapid expansion brings with it. When dozens of universities improvise an AI major in a single academic year, there is a temptation to strip it of real technical content and turn it into a fashionable label, with the risk of producing far more graduates than the governance and safety jobs the industry will actually create in the short term. It is a pattern already seen with other trendy disciplines: credential inflation is a real problem, not an anecdote.

In the long run, however, this current connects directly with the underlying thesis we defend at Zendoric: for AI to lead to abundance rather than a badly managed risk, what is needed is precisely a large, well-trained generation that understands these systems from the inside, not just those building them at full speed. That thousands of young people are choosing to study why models do what they do, instead of merely using them or fearing them, is one of the healthiest signs a society that takes both the opportunity and the risk seriously can give.