California has long been a leader in technological innovation, including the recent development of artificial intelligence (AI) and large language models (LLMs). But California residents are wondering how this new technology will impact their futures.
As Pomona College prepares to host a gubernatorial debate Tuesday, April 28, Pomona College President G. Gabrielle Starr examines the assumptions we make about AI, where human intelligence surpasses AI capabilities, and how college students can prepare themselves to thrive in a changing economy.
How does the human need for social learning limit AI’s ability to educate us?
Human beings gain knowledge by exploring the world, and we have a natural social desire to share that knowledge and exploration with other people. That desire is crucial: human interaction actually motivates us to learn. And that will remain a part of our humanity regardless of the technological frameworks that surround us.
Although humans can learn from anything, there are limits to what AI can help us learn. Interactions with chatbots and AI agents are superficial; they don’t provide the same sensory inputs that help us make meaning and develop knowledge from face-to-face conversations with other people. We’re also built to learn by trial and error. If you simply input a math problem into a program until it gives you a correct answer or tells you you’re right, you’re not learning from a teacher who can explain how you got the answer wrong. It’s the how that really helps you learn.
From your perspective as a neuroscientist, what is our riskiest assumption about AI’s capabilities?
In existing large language AI models, what we refer to as “intelligence” is actually prediction. AI analyzes words, diagrams and equations; it parses incoming data; and it sifts through and recombines existing knowledge based on probability and pattern. But even though AI is a technologically impressive tool, it’s ultimately a sophisticated prediction engine. It's not offering us knowledge, insight or meaningful connection.
Right now, many people are worried about a near-future where AI capabilities will surpass human ones, and what that means for our economy, creativity and humanity. There are real and understandable concerns here. But the real risk is that we’ll mistake AI for something it isn’t—and make policy decisions that impact our economy, creativity and humanity based on a faulty assumption.
What can human intelligence do that large language models (LLMs)—what we call generative AI—cannot?
So many things!
Artificial intelligence echoes computational structures in our brains, which is one of the reasons why one might imagine it’s comparable—or even superior to—human intelligence. But its outputs are derivative. Because it’s a prediction engine, it provides us with predictable responses, built on algorithmic optimization and limited by the scope and accuracy of what’s available in its training data.
Human intelligence, on the other hand, is social and expansive. Our minds thrive on surprise, ambiguity and the unexpected. We’re able to think critically, problem-solve and make judgments—especially when it comes to assessing whether AI outputs are correct.
Perhaps most importantly, human intelligence is innovative and collaborative. Our collective knowledge grows when we exchange ideas, grapple with challenges to our existing ways of thought and work together to tackle big problems. It’s that kind of collaborative innovation that helped humans invent AI in the first place: this technology is possible because of contributions from computer scientists, mathematicians, cognitive scientists, psychologists, linguists and so many others.
What skills and knowledge should college students build to help them compete in the current economy?
Our highest and best value—as students, employees and humans—is never going to lie in what we know. Our value lies in what we do with our knowledge, our ethical actions in the world and our creative potential.
Leaders are increasingly recognizing this. Since they already assume recent graduates know how to craft prompts, leaders are instead looking to hire employees whose skills go beyond technical knowledge and AI fluency, employees who can redefine problems, reimagine possibilities, negotiate conflicts successfully and make decisions in uncertain situations.
The most important skills that college students can develop are critical thinking, judgment and an ethical lens—skills that AI lacks. If students can strengthen their knowledge and experience in these areas, they'll demonstrate their value to employers.
How is Pomona College preparing students to succeed in an AI workforce?
We’re a liberal arts college, so we focus on broad intellectual development rather than specific career preparation. Pomona students engage with a variety of academic disciplines, with opportunities to conduct hands-on research alongside world-class faculty, and we offer courses and majors that combine traditional fields of knowledge in new ways.
Recently, we’ve seen increased student interest in fields that incorporate social sciences with policy and ethics. I believe our students are drawn to this well-rounded, liberal arts approach to the world precisely because it teaches the kind of human-centered problem-solving that AI can’t reproduce, and guides graduates from all disciplines—including scientists and technologists—to make more humane decisions. Ultimately, it prepares them to contribute to the world as leaders, scholars, artists and engaged members of society.
And we have proof that the liberal arts provides students with the skills they need to succeed in a contracting economy. At Pomona, 92% of our graduates secure employment or begin postgraduate education within six months. It’s because we’re educating students not only on what AI can do, but what it can’t: be creative, collaborative and open-minded.