College students are receiving mixed messages about the role of artificial intelligence in their education. EDUCAUSE Senior Researcher Nicole Muscanell said that many of them are told not to use AI in their courses, while employers are asking for graduates to come AI-ready into the workforce.
In EDUCAUSE’s 2026 Students and Technology Report, 53 percent of students said their courses had restrictive AI policies. Meanwhile, AI and big data are among the fastest-growing skills employers expect to need through 2030, according to the World Economic Forum’s Future of Jobs Report in 2025.
To help bridge this gap, Muscanell said colleges and universities should be focusing on a few baseline skills.
“You need to have a base-level knowledge,” she said.
The necessary AI skill set will vary from one profession to another, she said. A data scientist may need advanced AI capabilities, while a healthcare professional may primarily need to understand how to use AI-enabled tools and interpret their outputs. Muscanell said most fields will require training more in line with the latter: understanding a typical user interface and evaluating AI outputs.
She cautioned that higher education leaders should not overcommit to restructuring education around AI. Even as AI is reinvigorating longstanding questions about whether traditional assessments like multiple-choice tests adequately measure student learning and workplace readiness, she said higher education leaders should zoom out on the structural problems that come with managing change, rather than focusing on the specifics of this particular technology.
It is difficult for institutions to navigate rapid technological change with limited resources, she said, pointing to staffing shortages and expanding workloads as barriers to proactive planning at the institutional level. At the instructor level, faculty need time to understand emerging tools and determine how to incorporate them into teaching.
“Disruption and change are just going to keep coming at us,” she said. “AI is one of those disruptions, and so I think institutions need to get better at finding space to do more of the preparation and proactive stuff.”
Video Transcript:
It’s a confusing world right now for students because what they’re hearing is, for your career and to enter the workforce, you need to learn AI.
Then they’re showing up to their courses and they’re getting mixed policies and mixed guidance. And then even for the ones where they’re encouraged to use it, there’s not a whole lot of instructors having thoughtfully integrated AI into the course and the assignments and the assessments. And I think this is more or less because faculty also just haven’t had the time to fully grasp the tool themselves.
Even beyond AI, people for a long time have been asking the question: Are the assessments, like multiple-choice tests and the things that we have students do in college courses, are these the right assessments when it comes to making sure that students are learning and that they’re prepared for the workforce when they leave college? AI has jumped in and now it’s like, okay, it’s making it harder to assess those things.
Absolutely we can probably all agree, like, will you need to know some AI things for your job? Probably yes, and probably for many careers.
I do think that probably for many fields, the goal is that these tools become really user-friendly so that people don’t have to suddenly become experts.
For example, we know it’s going to impact the medical field. It already is impacting the medical field. But I think it would be so unreasonable to say, OK, You’ve gone to medical school for all these years and now guess what? You need like, a Ph.D. level in AI as well.
There will probably be like, you need to have a base-level knowledge where you are at least familiar with coming in and using something that has a user interface where it does have like, menus and things that you can toggle between and you have to know how to read certain outputs. But maybe you don’t have to actually understand the science behind like, a large language model necessarily.
Leave a Reply