By Josh Napier, College of Arts and Sciences
One day soon, if it hasn’t happened already, a hospital near you will switch on an AI system that recommends treatments. The AI will read charts, compare patients across thousands of similar cases and suggest what a clinician should do next. Whether any of it improves care depends on the people in the room, and on how well they and the system understand each other.
That hospital is one scene in a much larger shift, surfacing everywhere we live, learn and work. In May 2026, Google rebuilt its search box around AI, calling it the biggest upgrade to search in more than 25 years. AI now also helps screen job applications, flag fraud, tutor students, write software and translate language, often inside tools people don’t think of as AI at all. And every one of those systems has a second half.
The National Institute of Standards and Technology describes AI systems as inherently socio-technical, meaning their outcomes are shaped by societal dynamics and human behavior, not by code alone. However capable the model, its value depends on the person who frames the question, reads the answer, calibrates their trust and decides what happens next. Machine capability meets human attention, language, trust and judgment in every single interaction.
That is why understanding intelligence, human and machine, is becoming essential rather than optional, both for the people who design these systems and for everyone who works and lives alongside them. When machines participate in decisions, what should we understand about the people who interpret, trust, resist and sometimes defer to them?
This is where cognitive science comes in. And this fall, it comes to Florida. The USF College of Arts and Sciences welcomes the first students into a new Cognitive Science major, the first program of its kind in the state.
What is cognitive science?
Cognitive science is the interdisciplinary study of the mind and intelligence, and it has been asking these questions for decades. Researchers in psychology, neuroscience, computer science, philosophy, linguistics and mathematics have mapped how people perceive, learn, remember, reason, decide and use language. Many of the ideas behind modern AI grew from that longer effort to model the mind, even though today’s systems operate very differently from human ones.
The questions are not new. Their reach, from the research lab to the modern office and classroom, is.
The human problem inside the technology
Now go back to that hospital.
A clinician must know when to trust the recommendation and when to question it. The interface needs to communicate uncertainty without overwhelming the person using it. The system must account for bias in its training data. A patient must then be able to understand what role the technology played in a life-altering decision. And, in the end, someone must remain accountable for the result.
“The goal shouldn’t be to make people trust AI as much as possible; it should be to
help them trust it the right amount,” said Mark Pezzo, associate professor of psychology.
"That means knowing how accurate the system is for a particular task and communicating
uncertainty in ways people can use. But accuracy isn’t the only thing people care
about. In our research on medical decision-making, people were willing to accept a
small reduction in diagnostic accuracy when an AI system could provide an immediate
answer rather than make them wait for a human physician.”
None of those demands can be met by computational performance alone. They depend on
how people perceive information, form mental models, weigh evidence and respond to
authority. The same pattern appears in education, product design, cybersecurity, law,
public policy and management. Each challenge is simultaneously technical and human,
and people who can look at both halves of the problem with equal fluency are scarce.
AI is also changing where human expertise creates value. For much of the digital era, education and industry rewarded the ability to find information, perform an analysis, write code and turn knowledge into polished work. Generative AI now completes portions of that process at remarkable speed. Competent output is becoming abundant. Sound judgment is not.
A fluent memo isn’t necessarily a good decision, because a model can summarize the available evidence without recognizing what’s missing. It can recommend an action without understanding the people affected by it, and it bears no legal or ethical responsibility for the outcome. So human expertise moves upstream: toward defining the problem, choosing the goal, testing the assumptions and deciding when an answer shouldn’t be accepted. These are sometimes called “soft” skills, as though they sit apart from technical skill. In practice, they’re what make technical skill useful.
Anyone who has ever defended a humanities degree at a dinner party may permit themselves a small smile here, because the point is now being made well beyond this college. According to Nils Gilman, a former associate chancellor of UC Berkeley, when machines can generate plausible work on demand, universities must become more deliberate about cultivating judgment, goal-setting, live reasoning, trust and the willingness to revise one’s thinking. The point isn’t that older forms of learning suddenly became relevant because of AI. It’s that AI made their relevance harder to ignore. A university’s task isn’t to compete with machines at producing first drafts. It’s to help students develop the knowledge and judgment to decide which questions are worth asking, which answers are credible and what should happen next.
We’ve been making a version of that argument for a long time. Now the labor market is making it with us.
The U.S. Bureau of Labor Statistics projects computer and mathematical occupations to grow 10.1% from 2024 to 2034, compared with 3.1% across all occupations, and identifies AI adoption as a major source of demand. A market assessment commissioned by USF found more than 4,000 entry-level postings over a recent 12-month period connected to cognitive-science backgrounds, in roles running from human-computer interaction and user-experience research to data science, product management, behavioral analytics and AI governance. The titles vary, but the roles meet one basic need: people who can navigate both the logistics needed for developing technical systems and the emotional intelligence to understand what motivates human beings.
No single discipline is enough
Cognitive science is interdisciplinary by necessity. Intelligence cannot be adequately explained from one vantage point.
Computer science can show how a model processes information. Mathematics and statistics can test whether its results hold up. Psychology can reveal how a person interprets its output. Linguistics can explain how meaning changes through language. Neuroscience can examine the biological systems that make human cognition possible. Philosophy can ask whether the goal being optimized is the goal we should pursue at all.
Those perspectives don’t merely sit beside one another. A student trained across these fields can ask a fuller set of questions:
- How does a person form a mental model of an AI system?
- When does automation improve a decision, and when does it encourage overreliance?
- How should an interface account for the limits of attention and memory?
- What’s the difference between recognizing a pattern and understanding it?
- What does meaningful human oversight require?
That breadth prepares students for a labor market no curriculum can predict. Tools will change. Job titles will change. Some tasks students practice today may be automated before they reach the midpoint of their careers. What will endure, however, is the ability to learn across domains and evaluate unfamiliar systems.
How USF built it
USF has the pieces and is compiling them into something rare: a major that examines intelligence from both sides, human and machine.
Across the College of Arts and Sciences, psychologists study how people think and make decisions. Linguists examine how meaning moves through language. Philosophers ask what counts as knowledge, intelligence and responsible action. Mathematicians test the models used to explain and predict behavior.
Elsewhere at USF, the Bellini College of Artificial Intelligence, Cybersecurity and Computing brings expertise in the computational systems increasingly performing work associated with those human abilities. The two colleges are already building together. This Fall term they also launch a joint bachelor’s degree in computer science and interdisciplinary social sciences, founded on the same conviction that better technologists understand the people their work affects. The new Cognitive Science major extends that partnership, putting both colleges’ strengths into conversation around questions neither side can answer alone.
The curriculum follows the same logic. Students will move among computation and AI, psychology and linguistics, philosophy and the humanistic study of reasoning, mind and language, and mathematics and statistics. They won’t have to choose between understanding people and understanding technology. They’ll study both, and more importantly, how the two shape one another.
“USF is unusually well positioned to study intelligence from both sides,” said Elizabeth Spiller, dean of the USF College of Arts and Sciences. “Our college has spent decades studying how people perceive, reason and decide, while the Bellini College is advancing systems that can perform work once associated only with human intelligence. Cognitive science brings those conversations together. Students who can hold both in view won’t just be ready for the jobs this decade creates. They’ll help decide what those jobs should be.”
No program can tell students what AI will look like 10 years from now. What it can do is teach them to ask better questions as the systems change. AI will keep getting better at producing answers. Higher education’s responsibility is to prepare people who know what to do with them.