David Laszczkowski
Co-Founder, EdPilot
Understanding how to work with AI — not just use it, but understand its limits, evaluate its outputs, and apply it appropriately — is quickly becoming table stakes for professional competence in most fields.
Twenty years ago, universities invested significant effort in teaching students how to evaluate sources — how to distinguish a peer-reviewed journal from an opinion site, how to identify bias in primary sources, how to construct a research trail that holds up to scrutiny. This was considered a core skill, not a specialty.
AI literacy is in the same position now. Understanding how to work with AI — not just use it, but understand its limits, evaluate its outputs, and apply it appropriately — is quickly becoming table stakes for professional competence in most fields. Most university programs aren't teaching it yet.
The term gets used loosely, so it's worth being specific. AI literacy isn't about being able to write a good prompt, though that's part of it. The more substantive components are these:
Understanding what AI systems can and can't do. Current AI systems are remarkably capable at certain tasks and reliably poor at others. Students who understand the difference use these tools more effectively and catch errors that students who don't understand the difference miss entirely.
Evaluating AI outputs critically. An AI-generated answer that sounds authoritative may be subtly wrong, partially fabricated, or accurate but incomplete for the specific purpose at hand. The skill of checking AI outputs — against primary sources, against domain knowledge, against the specific requirements of the task — is one that needs to be deliberately developed.
Understanding context and boundaries. AI trained on general data performs differently than AI trained on specialized materials. AI optimized for fluency performs differently than AI designed for accuracy. Students who understand these distinctions can select the right tool for the task and set appropriate expectations.
Thinking about attribution and transparency. When and how to disclose AI use, how to cite AI-assisted work, and how to think about intellectual ownership in an AI-assisted context are all questions that students will face in professional settings and need frameworks for.
There are a few reasons this hasn't moved faster.
One is velocity. AI capabilities have developed faster than curriculum review cycles. By the time a course is proposed, approved, designed, and staffed, the landscape has shifted.
Another is disciplinary fragmentation. AI literacy is genuinely cross-disciplinary — it matters differently in law than in biology, in journalism than in engineering — but the mechanisms for building cross-disciplinary curricula are slow.
The third reason is that the faculty who would design these courses are still developing their own understanding. You can't teach something you haven't worked through yourself.
The full curriculum solution takes time. In the meantime, the most tractable approach is integration rather than standalone courses.
Faculty who are using AI tools in their own research can incorporate reflection on those tools into their teaching. Courses that involve research or writing can build in explicit discussion of when AI use is appropriate, what its limits are in that context, and how to verify AI-assisted work. Study environments that use AI can structure student interactions to build critical engagement rather than passive consumption.
None of this requires a new major or a major curriculum revision. It requires faculty who are thinking carefully about how they're preparing students for a world in which AI is a constant in professional and intellectual work — and building that preparation into what they already do.