On Monday, over 200 researchers and economists, including 15 Nobel laureates, released a joint statement pleading with government and industry to address the economic and social impacts of artificial intelligence, based on the belief that AI will result in the displacement of vast stretches of the workforce. Industry itself is plowing ahead with vast investments in AI infrastructure; Meta just announced a massive expansion of a data center being built in Louisiana, raising the cost to $50 billion. However, this seems a tad optimistic, given it represents a technology that most people don’t want and no one asked for, while making our entire society collectively dumber.
One of my favorite dystopian short stories is “Harrison Bergeron” by Kurt Vonnegut. Set it 2081, the story describes a government that enforces absolute equality by forcing citizens with above-average strength, intelligence, or beauty to wear debilitating handicaps. This world where the government handicaps the gifted leads to societal stagnation, where agents of the government resort to drastic measures to enforce their brand of “equality.”
Today, we are seeing a similar scenario play out, but with AI. Instead of the government handicapping the most intelligent people to make them seem less intelligent, society is becoming ever more dominated by a tool that makes less intelligent people appear smarter, while still making society dumber overall. AI represents a different mechanism and means, but it achieves the same basic societal effect: intellectual and artistic stagnation and decay.
The most obvious example of how AI destroys the human ability to learn and think critically is its ubiquitous use in high schools and colleges. A study released in October 2025 found that 84 percent of high school students use AI to brainstorm ideas, edit or revise essays, and/or conduct research and find sources. Another 69 percent use ChatGPT to help with school assignments and homework.
At the college level, the picture is no better: 85 percent use it for the basics of research and brainstorming. Another 57 percent of college students use AI at least weekly for coursework, and 46 percent say it is “extremely important” to use it in order to understand complex material, save time, and get better grades. Male students are more likely to use it than female students: What normal 19 year old wouldn’t rather be playing beer pong than reading Immanuel Kant?
AI’s allure is understandable. I took Philosophy 101 in the fall of 1994 at the Naval Academy before we had the internet, and I found Kant’s writings to be so dense and impenetrable that I absorbed almost nothing from my hours of reading his work. An AI summary would have been incredibly useful.
But would it have improved my comprehension? The research says no.
A study by MIT found that students who use AI to write show far less brain activity than those who used classic Google searching (with AI off) or those who used neither. Other studies find that students who use AI retain far less of the information that ends up in their writing, possibly as a result of “cognitive offloading” and “cognitive surrender.” Then there are also the issues with AI hallucinating information and citations, or being confidently wrong, which feeds misinformation to students who have never been taught to “trust but verify.”
From my own perspective as a writer, I know firsthand how easy it is to fall into these traps. Google AI search results frequently misattribute or misrepresent the contents of an article. AI-generated slop filling my social media feed often contains wrong information, which I wouldn’t catch if I wasn’t an expert in the areas AI was talking about due to years of reading primary source material. One of the funniest (and worst) examples of this happened when an AI search tried to confidently gaslight me, saying that something never happened, when I was physically there to observe it 30 years ago.
Other times, it generates plausible-sounding stories and videos, but with something about them just slightly … off. The sound. The setup. The people’s behavior. The narrative doesn’t seem quite plausible. But for people—especially younger people—who never had a chance to develop the experience and critical reasoning skills to constantly question what they see and hear, this artificial content will often go unnoticed.
At best, we’re talking about harmless fakes of doorbell cameras recording animals doing silly things. At worst, they are used to create self-reinforcing narratives about what sort of people don’t belong in our society. AI videos and algorithms feed our worst fears, desires, impulses, and beliefs as a culture, in much the same way that The Protocols of the Elders of Zion fed the worst in human gullibility over 100 years ago.
Like in “Harrison Bergeron,” government and industry are colluding to foist this on all of us whether we want it or not. They convince untalented, lazy, or unintelligent people that with AI, they can be just as talented as people who spent decades perfecting their craft. You can make movies! Write poems! Create “art”! Just give the algorithm a series of prompts, et voilà! Anyone can be an artist/moviemaker/writer/poet.
Unfortunately, what comes out the other end of the AI machine is almost universally awful and derivative. While over half the new e-books on Amazon are AI-generated, they generate fewer reviews (most of which are bad anyway) and low sales. Fundamentally, AI is uncreative: It takes all the information available and uses algorithms to generate what it thinks will come next, like a Chinese room thought experiment run amok that uses 4.4 percent (and rising quickly) of all energy in the United States, and up to 90 percent of the total computing power available in the country.
Now that AI has gobbled up everything and started dumping slop everywhere, it is consuming its own product like a bad drug dealer. When AI systems fold AI products into the dataset they draw from, they reinforce and pollute their own algorithms, creating a self-licking ice cream cone. Over time, the data set that AI builds its algorithms on becomes increasingly polluted with its own slop, eventually resulting in model collapse. In laymen’s terms, it is like the children’s game of telephone, or making a Xerox copy of a copy of a copy: The output just keeps getting worse over time.
Despite all of this, industry and the government have formed an unholy alliance to make everyone use AI for things that it probably shouldn’t be used for. The Trump administration is “focused on accelerating federal AI adoption to improve execution, competitiveness, and U.S. strategic advantage,” according to the Brookings Institution, with lax oversight. The lethal strike on a girls school in Iran has been attributed to the military outsourcing targeteering work to Anthropic’s Claude AI model.
Before I left the RAND Corporation, there was a heavy push from leadership to use AI to write reports so we could brand the product as “AI-informed.” This seemed like a terrible idea at a think tank that is supposed to generate new ideas, knowing that AI mostly just recycles and reassembles old ones. Despite this, AI has become the new buzzword, like Lean Six-Sigma or synergy. The Ford Motor Company drank the whole pitcher of Kool-Aid, firing hundreds of employees in the belief that AI could do the job better than humans, only to lose billions and eventually rehire them when it discovered the limits of the AI propaganda it’d been sold.
The promise of AI was that it would replace people or at least make their lives easier. Neither is generally true: It just forces people to multitask and churn out more. The only field where people are being replaced wholesale is in computer programming, where AI is making leaps in capability. The problem is that an entire generation of new coders is being lost. A Fortune-500 tech company senior architect whom I spoke to off the record believes that they’ll remain, if only to vibe-code and manage with AI.
They foresee an Idiocracy-like future where technology is treated like magic. Society survives entirely through extreme automation, but the everyday population has completely lost the understanding of how or why anything works, relying purely on habit—i.e., AI generates mountains of code, but no one is left who can understand it.
There is a rare consensus between Democratic and Republican voters that they dislike data centers using up their water, creating heat domes and noise pollution in their neighborhoods, raising electricity rates, and making it more expensive to buy a computer. There also isn’t a great deal of demand for many of the products AI is offering (like books and movies), leading many expert observers to believe that AI is an investment bubble that will eventually pop.
Most people know about the “faster, cheaper, better—pick two” truism. AI, in both the creative and industrial/engineering worlds, provides such an unbelievable boost to the “faster” and “cheaper (at least for now)” attributes that people are willing to sacrifice quality. Many of them don’t think of the degradation in quality as substantial and consider it a sacrifice worth making. Most people agree that past inventions, in retrospect, were also worth this sacrifice: The printing press eliminated the huge amounts of information memorized by scholars and professionals, and the calculator degraded the human knowledge of the fundamentals of math that the slide rule required. The improvement in speed and cost on one hand, and the sacrifice of quality and human ability on the other, is a level of magnitude above these previous inventions.
However, the fact that so many students use AI suggests that at least that part of it is here to stay. AI pollutes our creative and information ecosystem by making seductively cheap dreck. As in Vonnegut’s story, it makes the dumbest seem smarter, the smartest seem dumber, while dragging the entire human average down in the process.
