When AI Makes Intelligence a Commodity
We had a good run...
Image: Depositphotos
Source: Tom Davenport, PhD
For hundreds if not thousands of years, human intelligence has been a rare and valuable good. Those who had large amounts of intelligence could monetize and achieve high social status with it. Political leaders sought them out. Universities paid them to teach, do research, and publish on the topics about which they were smart. Being a “knowledge worker”—one whose job involved the creation, synthesis, or distribution of intelligence—was a role to which many aspired, and was often viewed as a credential leading to long-term and lucrative employment.
Companies and organizations are also partially built upon rare intelligence. They have specialized technology, processes, and technology, all of which they treat as “intellectual property” and which came from their knowledge worker employees. More broadly, the idea of business functions—finance, marketing, HR, etc.—is based on the intelligence needed to operate those parts of the business. If I’m a marketer, I should have intelligence on topics like the 4Ps (product, price, place, promotion). To learn those skills, most businesspeople went to universities, with undergraduate majors in business or an MBA. If you wanted legal or medical knowledge, you went to graduate school for degrees in those fields.
Those of us who create and apply intelligence for our livings have had a great run, but our status and livelihoods are now at risk. The culprit, of course, is AI—specifically generative AI. I fear that learning, creating, communicating, and debating intelligence are increasingly activities that genAI can do better than the great majority of humans. I don’t want to get into the issue of whether we have achieved AGI or artificial general intelligence yet, but we are dangerously close to it. And the technology is already commoditizing intelligence that lives outside our brains and perhaps giving us less motivation to add it to our own personal stock of smarts.
Commoditized Intelligence for This Professor and Pontificator
I’ve been teaching an AI course at Babson since 2015, and I am definitely seeing signs that for students genAI is making intelligence a commodity. Of course I allow and even insist on AI use for assignments—how could I not do that in a course on AI.? The assignments generally take the form of essays, presentations, or videos. All of these content buckets can now be created by AI, and it does a good job of finding and packaging the intelligence in each format. I thought I had essays figured out; I ask the students to describe their prompts, make visible edits to the LLM output, and verify all citations. Although the students really don’t like those tasks because they lower their productivity, I do think tjeu engages their brains somewhat.
However, this academic year I have noticed that student presentations are: a) very easy to create with minimal prompting; b) pretty high quality; and c) difficult to assess in terms of the relative contributions of the student vs. AI. That’s also true of the videos they use AI to create. I often feel that I am grading the AI rather than the student. I ask the students to send me the original AI-created versions of their presentations, and then the versions they edited. For the most part the changes they make are superficial, and I even ran both original and edited versions through Claude for a couple of students to confirm my suspicions. Claude agreed with me that most of the big ideas came from him/it.
Perhaps driven slightly crazy by grading my students’ AI-created content, I decided to experiment with my own. I had agreed to give a presentation next month on a topic I had never previously addressed but was generally familiar with. So I asked Claude to give the presentation a shot. It did a pretty amazing job on a first draft with only a minimal prompt. I immediately edited the output somewhat, but later I realized that the same criticism I was leveling at my students applied to me as well: most of the big ideas came from Claude.
I decided to re-do the presentation, identifying the ideas that were AI-created and creating some new ones of my own. Who knows if it’s any better than the first version, but at least I will have engaged my brain in the topic. But if I had used the Claude output, the distinguished audience for my presentation would probably have thought it was fine. They would have traveled many miles to a conference, only to see a person deliver a presentation that was largely created by a machine. If they realized it, the next time they might just stay at home and ask Claude what it thought about the issue. Perhaps conferences are among the institutions that will slowly disappear.
What Does Commoditized Intelligence Look Like?
From my students’ behaviors I can begin to see what a commoditized intelligence world might look and feel like. Perhaps no one would bother reading any long-form content anymore, because AI could easily summarize it for them. Most authors may eventually succumb to the temptation of having AI do their writing for them. In other words, the books and articles that continue to exist will be written by AI and read by AI.
Consultants, attorneys, doctors, professors, and other knowledge workers will no longer be needed—at least not for their knowledge—as their level of intelligence is easily available through a few prompts. Entertainment and cultural products of various types will also be created by machines, because they will know what people find beautiful, emotionally resonant, or amusing. And while DOGE didn’t do much to dismantle the government bureaucracy, AI might succeed where it failed for government employees who were hired for their expertise.
Will universities continue to exist? I certainly hope so, but I have my doubts. In the Boston Globe today there is an article entitled, “What’s the Point of MIT?” It focuses particularly on math Ph.D. students, who were gobsmacked by all the unsolved math problems for which OpenAI has been able to create proofs. In a small survey of the math grad students for that article, 71% said that they are planning to abandon academic math careers. Certainly colleges serve other functions than research—teaching, for example—but AI is also getting increasingly good at personalized teaching and tutoring. Maybe there will still be a need for institutions to get teenagers out of their family homes and to teach them some interpersonal skills, but that kind of institution would probably be very different from today’s university.
What Kinds of Intelligence Will We Still Need?
Emotional intelligence is a good example of the type of capability that humans may need to focus upon when AI takes over what Howard Gardner, the father of “multiple intelligences,” referred to as “linguistic”and “logical-mathematical” intelligences. It’s also quite possible that AI and related technologies will soon surpass us in a few more of his types, i.e., visual-spatial, musical, and bodily-kinesthetic, although that may take a while. Given how well some apps already recognize plants and animals, “naturalistic” intelligence—a more recent classification advocated by Gardner--may also succumb to AI.
It’s looking like we’ll have plenty of time to focus on each other and ourselves. Gardner broke down emotional intelligence into “interpersonal” and “intrapersonal,” distinguishing between understanding what other people are thinking and feeling, and navel-gazing into one’s own mind and heart—and navel, presumably. I don’t doubt that AI can already help with those—lots of people, for better or worse, are using AI for therapy these days and using chatbots to rehearse difficult conversations—but there is still an important human role to be played in these intelligence domains.
But in our AI-dominated future, perhaps the single most important aspect of human intelligence in Gardner’s list is “existential,” or the “tendency to ponder deep philosophical questions about human existence and meaning.” I’m guessing we’ll need a lot of that, and we’ll have plenty of time to cultivate that form of intelligence.
Of course, there are a lot of smart people in the world, and perhaps they will figure out something better to focus their brains on than understanding each other and themselves. I’m no fan of Elon Musk, but he may be onto something with Neuralink. It could be that the only way to ensure an important role for human brains is to augment them surgically with machine intelligence.
Tom Davenport, PhD
Tom Davenport is a world-renowned thought leader and author, is the President’s Distinguished Professor of Information Technology and Management at Babson College, a Fellow of the MIT Center for Digital Business, and an independent senior advisor to Deloitte's Chief Data and Analytics Officer Program.
An author and co-author of 25 books and more than 300 articles, he helps organizations to transform their management practices in digital business domains such as artificial intelligence, analytics, information and knowledge management, process management, and enterprise systems.
He's been named:
A "Top Ten Voice in Tech" on LinkedIn in 2018
The #1 voice on LinkedIn among the "Top Ten Voices in Education 2016"
One of the top 50 business school professors in the world in 2012 by Fortune magazine
One of the 100 most influential people in the technology industry in 2007 by Ziff-Davis
The third most important business/technology analyst in the world in 2005 by Optimize magazine
One of the top 25 consultants in the world in 2003
For more information: http://www.tomdavenport.com