Let me start with a number that stopped me in my tracks.

This year, MIT surveyed its own students, faculty and staff about artificial intelligence. Nearly 80% said AI skills will be necessary for their future careers. But when undergraduates were asked whether MIT was actually preparing them to use AI professionally, only 25% said yes.

Twenty-five percent. At MIT.

These are students with every advantage you can think of. World-class faculty. Unlimited access to the newest tools. Classmates who build AI for fun on the weekends. And still, three out of four don't feel ready to use it on the job.

So here's the question I keep coming back to. If they don't feel ready, what about the first-generation student juggling classes and a part-time job? The adult learner retraining for a new career? The high schooler who will be the first in their family to walk into a professional office?

Source: MIT Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training, 2026

So what did MIT actually find?

Back in January, MIT pulled together a committee of students, faculty and staff and gave them a big job: figure out how AI is really being used on campus, where it helps, where it hurts, and what the school should do about it. They released their final report in August after months of work.

Some of it is what you'd expect. Most people on campus use AI every day and feel like it makes them more efficient. But some of it is pretty sobering.

The committee found that AI can now produce believable answers to almost any written assignment, from essays to math proofs to code. They noticed students skipping office hours and studying together less. And they warned that when students grab AI the moment something gets hard, it can create what they called "the illusion of learning." You feel like you got it. You didn't.

To MIT's credit, the report doesn't say "ban it." It calls for rethinking how teaching, assignments and grading work so that AI supports real learning instead of quietly replacing it. We've made the same case for faculty AI adoption and student AI readiness in teacher preparation: one without the other doesn't get students where they need to be.

Access was never the problem

Here's what jumps out to me. MIT students aren't short on AI. They use it constantly. And yet most of them still don't feel ready for work. It isn't just an MIT thing, either. When we looked at the class of 2025, half of new graduates said they felt unprepared for AI at work.

That tells us something important. Using AI to get through homework and knowing how to use it well in front of a boss, a client or a patient are very different skills. The second one takes judgment. When should you trust the output? When should you push back? What should never go into a chatbot in the first place? How do you explain your reasoning when a decision matters?

That kind of judgment is the heart of AI literacy, and nobody builds it by accident. It comes from practice, in situations that feel real, with someone helping you think through the tradeoffs.

The gap gets wider for students who start with less

Students at well-resourced schools often pick up workplace know-how from internships, alumni networks and parents who use these tools at their own jobs. Plenty of the students I work with don't have that. Nobody at the dinner table is talking about how their team uses AI to draft reports or flag errors.

That has nothing to do with talent. These students are sharp, hungry and more than capable. It's about exposure. And if even MIT's students are saying "I'm not ready," then programs serving everyone else can't assume students will just figure it out on their own. We have to teach AI literacy on purpose.

What real AI literacy looks like

So if access isn't enough, what is? From our work with schools, colleges and workforce programs across the country, here's what actually moves the needle.

1. Practice in situations that look like real jobs

   Students step into realistic scenarios as a nurse, a paralegal, a marketing coordinator or an operations lead, and they use AI to get real work done. Then they make the call.

2. Keep the healthy struggle in the room

   MIT's big worry is students letting AI do the thinking. The fix is designing tasks where students have to question the output, catch the mistakes and defend their decisions. AI becomes a thinking partner, not a shortcut.

3. Put ethics first, not last

   Bias, privacy, accuracy and knowing when not to use AI at all. These aren't extras to cover at the end of the semester. They're the whole foundation.

4. Give students proof they can do it

   A certificate of completion is nice. A portfolio of real work that shows an employer exactly how a student used AI responsibly is a whole lot better.

The good news

MIT framed its report as a call to action, and I think that's exactly right. But you don't need MIT's budget or MIT's engineers to answer that call. You need a clear plan, the right kind of practice and a commitment to putting human judgment at the center.

That's what we built AI Essentials to do. Our AI literacy certifications drop students into hands-on simulations with real AI tools, teach ethics from day one and leave them with portfolio work they can actually show an employer. No guesswork, no hoping they'll pick it up somewhere down the road.

The students who walk into interviews confident with AI a year from now will be the ones whose programs decided to act today. Let's make sure yours are part of that group.

Frequently asked questions