The accelerating ascent, ubiquity, and commercialization of artificial intelligence require a renewed focus on truly elite human capital if we are to safeguard the future of Western civilization — both from external adversaries like China and also, perhaps even more importantly, from ourselves, especially given our postmodern and transhumanist tendencies.

In the coming years, we will need an elite cadre of Americans residing at the top levels of national and state government and bureaucracy. And yet, we are confronted by a very sad state of affairs across K-12 and postsecondary education, making the creation of such an elite class an increasingly difficult task.

We are clearly sapping the attention spans and atrophying the brains of our high school students.

A recent Atlantic article illustrated “Exhibit A” of this problem, namely, Harvard, the peak of elite credentialing institutions. The article, titled “The Perverse Consequences of the Easy A,” documents an alarming trend after decades of grade inflation. This excerpt helps give a sense of the problem’s progression: “In 2011, 60% of all grades were in the A range (up from 33% in 1985). By the 2020-21 academic year, that share had risen to 79%.”

Harvard has studied the problem and its effects: It turns out that when little effort is required to succeed in traditional academic respects, students stop going to class and, unsurprisingly, are doing less and less learning. An embarrassing fact emerges from faculty and student interviews: Fewer students are reading books and engaging with ideas at the world’s leading bastion of higher education. Trends are similar across the Ivies. The rise of ChatGPT and other large language models only exacerbates the problem.

The collapse of true learning in higher education should not be a surprise: The supply side for higher ed — teenagers — are rapidly incorporating LLMs into their daily academic lives.

In January, a Pew Research survey found that the number of America’s teenagers, ages 13-17, using ChatGPT had doubled since 2023 from 13% to 26%. Awareness among teens of ChatGPT has grown significantly over the last two years as well from 67% to 79%. With increasing familiarity comes the rising likelihood of teens using ChatGPT for homework and paper writing, as well as the opinion that it is legitimate and good for such purposes — roughly 50% to 80% of those surveyed, depending on how familiar they are with the technology.

Some initial studies suggest that this problem may be worse than the rising temptation of machine-aided plagiarism. An MIT Media Lab study determined that the use of ChatGPT in researching and composing papers led to underperformance “at neural, linguistic, and behavioral levels.” The main author of the paper emphasized that “developing brains are at the highest risk.” The study is still under peer review and has a small sample size, but it would seem to confirm a common theme of similar cognitive and concentration studies done by many researchers since the rise of social media and the smartphone.

We are clearly sapping the attention spans and atrophying the brains of our high school students. The best of them are going to elite institutions of higher education, where they are less likely than ever to take any real advantage of their most important years for stocking intellectual capital and forming their minds and souls.

Technological Quixotism

Our pursuit of the holy grail of artificial general intelligence is sold to us by our current technologist class on at least two tracks. We are told that the AGI revolution will cure cancer, extend our lives considerably, help us terraform Mars, and usher in a new age of abundance and convenience. Who doesn’t want that? And we also really have to do it, pedal to the metal, in order to beat China in the new nuclear arms race — that is, the AI race.

This generally pro-technologist point of view was represented in the recent attempt by Sen.Ted Cruz (R-Texas) and others to get a 10-year moratorium on state regulation of AI into the One Big Beautiful Bill Act. That effort failed, thankfully, despite an intense lobbying effort by a growing constellation of pro-AI Big Tech PACs, super PACS, and lobbyists.

Another finding in the MIT study also lends credence to the recent enthusiastic embrace of AI. If you took the test group that was asked to complete a writing assignment without ChatGPT to rewrite their paper without it physically in front of them but with ChatGPT’s assistance, their measured brain activity demonstrated more robust engagement and retention, and the finished product was of good quality. This suggests that the use of LLMs as aids rather than originators of thought and writing posed much less of a probability of cognitive laziness and atrophy. In this way, LLMs look more like a useful supplemental tool.

RELATED: The AI takeover isn't coming — it's already here

Photo by BlackJack3D via Getty Images

Students face a great temptation to use this new technology as a pedagogical aid, as some elite universities like Duke are trying to integrate AI and LLMs into their systems and educational strategies. But growing research suggests that doing so has as many dangers as advantages. Consequently, AI must be approached very cautiously.

Moreover, integration of LLMs into K-12 education is gaining steam, especially given the increasingly ideological bent of primary education in recent decades. If the education-school-credentialed leftists who disproportionately populate the ranks of our public and private K-12 teachers can’t be trusted, perhaps the solution is to cut them out altogether and replace them with AI.

The use of LLMs as aids rather than originators of thought and writing posed much less of a chance of cognitive laziness and atrophy.

This experiment is currently being run by the private K-12 Alpha School based in Austin, Texas. Alpha Schools now have 17 locations either starting or nearly ready to launch across the country, charging roughly $45,000 in tuition annually. They boast excellent results in testing metrics (SAT and ACT), even while offering only two hours a day of AI-tutor-based instruction, followed by another four to five hours (including lunch) of life skills and creative and collaborative group work under the guidance of real-life human mentorship.

This is a new experiment, so it remains to be seen how Alpha students will fare on a longitudinal basis as the first cohorts matriculate into higher education. The Alpha schools are relentlessly data- and testing-driven, so perhaps they will navigate this uncharted territory successfully, avoiding the pitfalls of screen-based learning and attendant tradeoffs.

A litany of pre-AI age studies show the positive benefits of students getting back to the basics of education before the introduction of the screen. Taking notes by hand leads to better retention and absorption of material compared to taking notes on computers, to cite just one example.

Don’t let your servant become your master

The larger looming problem, however, is how we should educate elite students — how we should cultivate elite human capital — and equip them to navigate a rapidly changing national and international technological environment that is still bedeviled by the perennial and ancient difficulties of preserving “small-r” republicanism and the common good.

The argument of our technological class is that elite students should be set free — and even subsidized and offered quasi-monopoly protection — to pursue the quest for AGI. If we don’t, they argue, we’ll lose the AI arms race, and the West will be eclipsed by China, militarily and economically.

To rip an international anecdote from recent headlines to illustrate our dilemma further, Russian President Vladimir Putin and Chinese President Xi Jinping were caught in a hot mic moment at a China confab discussing exciting advancements in biotechnology and organ harvesting — and even what such “advancements” might mean for their own longevity. If Putin is excited about living for another 20 to 50 years, Xi and his oligarchy must be pondering and planning for the possibilities of biotechnology, gene editing, eugenic embryo selection, and artificial wombs as a possible solution to China’s demographic problem.

Couple that impulse with the race for AI supremacy, and we must face the possibility — perhaps quite soon — of an arms race not only in AGI, but also onto transhuman vistas previously relegated to the pages and screens of science fiction.

Navigating this future while preserving America’s spirit of liberty and constitutionalism will be a tall order. It will require large bets on the old tools and contours of liberal education by private philanthropy and local, state, and national governments.

The ultimate control of our republican future must not be left to the technologists, but rather to statesmen and leaders whose minds and souls have been shaped in their formative years by a deep consideration of those age-old questions of justice, the common good, natural rights, human flourishing, philosophy, and theology.

The argument that we don’t have time will be a powerful one. The relentless pursuit of new areas of technical knowledge will be sold as the more urgent task — after all, national survival, they say, may be at stake. Given the 20th century’s experience with technical mastery severed from ethical, political, and constitutional safeguards, the bet on the unfettered pursuit of technological supremacy to the neglect of all else is just as likely to result in self-destruction.

As my colleague Christopher Caldwell has recommended, our AI arms race must be augmented, supplemented, and ultimately guided and controlled by wise statesmen who are steeped in the older ways of American liberal arts education. My hope is that those who are anxious about the fate of free government in the face of external material threats and internal spiritual threats can join forces to navigate our brave new world with wisdom and courage.

RELATED: AI is coming for your job, your voice ... and your worldview

Photo by Moor Studio via Getty Images

To that end, we urgently need to locate, recruit, equip, and refine as many members of America’s current and soon-to-be cognitive elite as we can find and help them become better readers, thinkers, and writers. They will then be properly prepared, at least to the extent we can help them to be, to balance our pursuit of technological progress — intelligently and humanely — with the traditions and principles of Western civilization.

We need a Manhattan Project for elite human capital. Our difficulty is that we can’t snap our fingers and replace the Harvards and Yales with Hillsdales. And yet something approximating that miraculous trick may be needed to save us from our international rivals — and from ourselves.

Editor’s note: This article is adapted from a speech delivered at the 2025 National Conservatism Conference. It was published originally at the American Mind.