The Boos Were Not About AI

This graduation season, something unusual kept happening at commencement ceremonies across America. Eric Schmidt, former CEO of Google, stood before ten thousand graduating students at the University of Arizona and told them that Artificial Intelligence (AI) "will touch every profession, every classroom, every hospital, every laboratory, every person and every relationship you have." The crowd booed. At Middle Tennessee State University, Scott Borchetta, the CEO of Big Machine Records, told media and entertainment graduates that AI is "rewriting production as we sit here." He got booed too and when he pushed back saying "deal with it, like I said, it's a tool", he got booed louder. At Glendale Community College, the college president tried to explain why some students' names were being misread or skipped entirely at the ceremony. "We're using a new AI system," she said. The crowd erupted. At the University of Central Florida, Gloria Caulfield, vice president of strategic alliances for the Orlando-based company Tavistock, called AI "the next industrial revolution." She seemed genuinely surprised when the boos hit her mid-sentence.

About seventy percent of college students, according to a 2025 Harvard Kennedy School poll, see AI as a threat to their job prospects. A recent Gallup survey found that among GenZ, anger about AI has risen while excitement and hopefulness have declined. However, as a Cornell graduate whose wife is currently studying at the Cornell Law School, and having been plugged into graduate and undergraduate communities, I can say for sure that these students who were booing are not Luddites. In fact, many of them are tech enthusiasts. Many of them use AI daily or weekly. So, the boos are not really about the technology. They are concerned about who is selling it, and what the track record of that group of elites looks like.

There was a time when Silicon Valley carried a different kind of reputation. When Facebook first rolled out beyond university campuses in the mid-2000s, it genuinely felt like something new. It was seen as a way to stay connected, to find people, to build things together. And it did, even from small town Firozabad in India, my father created a whole network of blood donors across many cities through Facebook groups. Similarly, Google's original promise was to organize the world's information and make it universally accessible. Twitter positioned itself as a global town square. The enthusiasm was real, and for a while, so was the trust. In 2017, nearly eighty percent of Facebook users agreed with the statement that Facebook was committed to protecting their personal information.

Then Cambridge Analytica happened. In 2018, it emerged that the political consulting firm had harvested the personal data of over fifty million Facebook users without their consent, and used it to build psychological profiles for political targeting. Trust in Facebook dropped from seventy-nine percent to twenty-seven percent in the space of a week. As one Ponemon Institute survey respondent put it at the time: "Facebook doesn't see any value in protecting the privacy of its users."

Overtime, this became a broader pattern that became clear across the decade. Social media platforms had been optimizing for engagement, which meant optimizing for outrage, anxiety, and compulsion. I remember visiting University of Virginia in 2017 in an editors’ conference representing Cornell Policy Review as its Editor-in-Chief where a PBS speaker claimed, through evidence, how bots and fake profiles were organizing anti-gun rallies and targeting those events to pro-gun population and that led to an actual pro-gun rally and this triggered an anti-gun rally. And in many places the opposite happened – a fake pro-gun rally spurring an actual anti-gun rally, which in turn spurred a real pro-gun rally. And social media business models amplified this.

This is the kind of inheritance that AI has received. The firms leading the AI wave are really the same firms or are led by a similar set of silicon valley elite who built the social media ecosystem, and they are now asking for a second round of trust. They are surprised to find there is not much trust left in them.

But here is where I think the graduating students, as legitimate as their anger is, are collapsing two separate things into one. The firms leading AI are less trusted but AI itself is not the same thing as the firms leading it.

AI comes with short-term, real consequential pains

My argument is that AI, if we look at it in an unbiased manner, isn't likely to reduce wages in the long run, but it will likely reduce jobs. I for one do not buy into the idea that AI will create a “useless class”, as Yuval Noah Harari would have you believe. This argument has a long and consistently embarrassing track record. In March 1812, a letter addressed to a shearing-frame owner in Yorkshire declared that the movement would "never lay down arms" until Parliament passed an act to put down "all machinery hurtful to commonality." Within a three-week period in Nottingham alone, over two hundred stocking frames were destroyed, and the government, fearing a national movement, positioned thousands of soldiers to defend factories and passed a measure making machine-breaking a capital offense. The Luddites were not irrational at all, interestingly. The wages of Nottingham weavers had fallen by an estimated minimum of thirty percent, and their fury was directed not at technology in the abstract but at a specific set of power relations surrounding it. What they could not see and what no one at the time could see, was that the same industrial transformation they were fighting would eventually generate more employment than it destroyed, and in entirely new categories of work that had no name yet. The same prediction of mass permanent uselessness followed electrification, then the automobile assembly line, then computing, then the internet.

Each wave produced its version of the Harari argument, dressed in the vocabulary of its era. Each time, human need and human ingenuity found new things to do. The fact of economy is that human needs are infinite, and resources are always scarce – so demand does not disappear. What the Luddites were actually protesting, if you read the letters carefully, was not that work would vanish but that the gains of new machinery were being captured entirely by the owners while the costs were being pushed entirely onto the workers. That is a legitimate grievance. It is also, notably, a distributional argument, not a uselessness argument. Harari collapses the two, which is intellectually convenient but historically sloppy.

Job reduction may not always be income evaporation

AI may not exactly evaporate jobs but it will reduce firm size. There is a fair possibility that it may expand the number of firms in the market as many firms outsource stuff to other firms. So, think of it as a world where each individual is like a firm and instead of a fleet of employees there are AI algorithms, machines, and some (maybe not in all businesses) robots. In that case, you can have a big economy, many companies or solopreneurs, but fewer "jobs". In fact, it is entirely possible that the solopreneurs may be earning a lot more than they would have if they were in a traditional job.

To understand how this may work, let us reflect on the question Ronald Coase, the Nobel winning economist, asked and it has always intrigued me. He asked if markets are so efficient, why do large firms exist at all? Why doesn't every task get contracted independently? His answer was that using markets has costs, i.e. search costs, negotiation costs, monitoring costs, enforcement costs. Firms emerge when it is cheaper to coordinate internally than to transact repeatedly through the market. But as firms grow, internal bureaucracy becomes costly too, so they expand only to the point where internal coordination is no longer cheaper than outsourcing.

Think again. In a pure market, supply exists and demand exists, and in theory they should find each other automatically. A farmer has tomatoes, a consumer wants tomatoes, a price appears, and the transaction happens. If markets worked perfectly, a large grocery chain such as a Walmart, a Wegmans, or Big Bazaar in India, would have no reason to exist in the form it does. It would simply be a shed, or at most a giant farmers market, where farmers show up with produce and buyers pick what they want. But the fact of the world is that they do and they have large procurement departments, legal departments, marketing and HR departments etc. Why do we see enormous stores with hundreds of employees, complex internal hierarchies, and entire departments dedicated not to selling food but to coordinating the process of getting food there in the first place? Coase's insight was that using the open market is itself costly. The farmer has to find buyers, the buyer has to find the farmer, someone has to verify quality, someone has to enforce the agreement if delivery fails, someone has to renegotiate when prices change. These search costs, negotiation costs, monitoring costs, and enforcement costs add up, and when they add up enough, it becomes cheaper to bring the coordination inside a firm rather than repeat it through the market every single day. It is cheaper to coordinate internally with the grocery chain's procurement officer, logistics coordinator, quality inspector, and contract because having them on salary, under one roof, subject to internal direction, is easier than going back to the open market for every shipment of every product from every supplier.

However, Coase observed that as the firm grows (or as its work in relation to the firm's capacities grow), its internal coordination becomes expensive in its own right, with more staff and bureaucracy slowing decisions and creating inefficiencies. So, at some point the firm stops expanding internal departments and hiring, and contracts functions out.

With AI, the procurement officer who spent her day calling twenty farmers to compare prices is replaced by a system that coordinates with all of them simultaneously and drafts a purchase order in minutes. This, like other internal managerial roles/departments, existed because internal coordination, however expensive, was still cheaper than the chaos of doing it through the open market repeatedly. AI changes this scenario by making market coordination easier through reducing search costs, getting more tailored solutions, and by automating a lot of decisions. And as it does this, the firm needs fewer internal staff to coordinate the same volume of activity. Functions that once required a full department can be handled by AI tools, and the grocery chain that employed five hundred people in coordination roles may find it only needs fifty.

At the same time, now the firm may want to do other things it wouldn’t have been doing earlier. The market’s evolution would necessarily mean new needs would crop up and to survive, firms would have to do a lot many things they do not specialize in. For that they would outsource and even the agencies they outsource too may be using AI, but it does create room for new kinds of markets and specialized services. Some examples, even though we may only find out exact examples later when markets evolve, could be recruiting a specialized marketing agency (which may use an AI of its own) to tailor customer experiences, or to reach new customers, or to do a whole range of things that the firm was earlier unable to do.

Essentially, AI has the promise to reduce the cost of search, contracting, communication, compliance, logistics, and design. When transaction costs fall, the rationale for bringing everything inside one large firm weakens. The economy may, in fact, become more efficient at the macro level with AI. Interfirm transactions may increase while intrafirm transactions might go down. Given that human needs are infinite there will still be a market. The challenge is that fewer of these needs may be met through the stable, long-term employment relationship that modern economies were built around.

The “job” as we know it today, comes with a lot more things than income. Currently in our society, especially in the USA, health insurance, retirement savings, social security, and for many visa status, is tied to a job. Even if income eventually doesn't disappear, AI will likely change the structure of the society where job dependence, as we know it, may vanish and people may find stability in having a firm that runs largely on AI and is protected by patents, trademarks, copyrights etc. There may be many such firms in the market. However, as a society we would need to tie our other structures such as health insurance etc. which have so far come as perks with a job, to these new entrepreneurial reality. For example, we might just need comparable health insurance coverage for smaller firms or solopreneurs. We may as well need an equivalent of an automatic provident fund saving scheme for solopreneurs.

Therefore, the situation that is unfolding before us deserves a serious institutional response, not a commencement speech telling graduates to "deal with it."

Hate for the player, not the game

At the same time, AI companies will gain disproportionate power and this will continue to worry us. But this too has precedent. There was a time when railroads were so central to economic life that their owners effectively had veto power over national logistics. Oil companies became synonymous with state interests across the twentieth century. Automobile manufacturers shaped cities, highways, tax codes, and foreign policy. Computer companies and then internet platforms went through the same arc. Each dominant technology produces a dominant sector, and that sector for a time becomes entangled with state power in ways that feel alarming. The question is not really about completely escaping what has always happened but adapting our socio-economic structure and institutions so that it doesn't happen in as unfair a manner as it had probably happened in the past.

The students booing at commencement are right to be skeptical. They have inherited a trust deficit. The answer to that is not in rejecting the technology and deep down, I feel nobody is rejecting the technology, at least in the student demographic. They are simply rejecting the players, not the game. The answer lies in institutional design.