Why institutions struggle with AI

I recently talked with Lance Eaton, Senior Associate Director of AI and Teaching & Learning at Northeastern University and writer of AI + Education = Simplified. We traded ideas about what’s actually working. We came up with 10 specific, practical ways anyone who teaches, coaches, or leads can put AI to work.

📺 Watch the full conversation above, or read highlights below.

10 Ways to Use AI 🛠️

Note: Lance and I alternated tips below 👇

1. Spark Richer Student Reflection 🪞

Lance: Ask students to reflect through a conversation with AI rather than staring at a blank page. A well-prompted AI will keep asking follow-up questions, pushing students past “I didn’t like it” toward real analysis.

2. Strengthen Your Syllabus 📋

Jeremy: Give an AI assistant your syllabus and ask for a critique — for clarity, inclusivity, student-friendliness, and completeness. You’ll get specific, honest feedback. The AI won’t write the syllabus for you, but it will challenge you to make yours better.

We don’t always have colleagues at our side who can offer input on our work. So this is an objective, independent, instant, constructive way to get a useful critique.

3. Make Materials More Visual 🎨

Lance: Turn your syllabus into a graphic version students actually want to read. AI assistants can help you create visual layouts and simple comics-style explanations without any design experience.

4. Improve Lesson Plans 📐

Jeremy: Describe your learning goals, your class size, your constraints — then ask AI to generate 10 warm-up or closing activities. You won’t use most of them, and you might remix a couple. But having options means you’ll often figure out something better than what you’d have designed alone.

5. Try It Until Something Clicks ⚡

Lance: Play with AI until it does something that genuinely surprises or excites you. That moment of “Wait, I could actually use this,” is what shifts the conversation from theoretical to real.

“For some students, this is really powerful, including students navigating English as a second language or ADHD or dyslexia — these tools can unlock things.”

Sponsored Message

Catch Up On Books: Maximize Your Time 📚

Life is short. Your reading list is probably long.

Shortform simplifies your journey to success. Whether you’re into self-improvement, business, or psychology, get in-depth book guides with summaries, actionable insights and exercises. Start applying what you learn immediately.

Each guide has a concise, useful one-page summary. You also get chapter breakdowns and practical exercises to apply what you’ve learned. Imagine mastering key lessons from books like Atomic Habits in just two hours. Accelerate your career and boost your productivity.

Stay ahead of the curve with new summaries added weekly, and use the AI browser extension to quickly summarize articles and videos, freeing up valuable time for what matters most.

Get a free trial and $50 off the annual plan.

6. Build Engaging Class Activities 🧩

Jeremy: When you need a compelling analogy for a hard concept, or a historical anecdote, or a mini case study for a short role-play exercise, AI assistants can be helpful in expanding what we consider. If you’re teaching a subject you know well, you can set the direction and take responsibility for verification.

NotebookLM and Claude can generate examples quickly, and can search your own notes to surface examples you’ve created yourself but lost track of. The goal of using AI in this context is strengthening engagement and improving the learning experience. It’s not for whiz-bang special effects.

7. Generate “Bad Examples” Safely 🚫

Lance: Examples can be useful to illustrate what not to do, but you’d never embarrass a student by presenting their work as an example of a mistake.

“We’re never, ever going to — nor should we — ask a student, ‘This was a really horrible thing, can I use it as a bad example going forward?’”

AI tools can generate intentionally flawed examples: a weak argument, a poorly structured paragraph, or circular reasoning. Students learn what to avoid.

8. Catch What You’re Missing 🔍

Jeremy: Ask an AI assistant to review your materials for accessibility gaps, unclear instructions or areas where your material could be more inclusive. Think of it as a thoughtful colleague who reviews your work and catches what familiarity made you miss.

9. Analyze Student Feedback 📊

Lance: Strip names and any identifying information from end-of-semester feedback, then ask AI to identify themes, patterns, and gaps. As Lance put it, “What are some things that I’m not seeing? What are some assumptions I’m making or missing? What are some ways I might redirect the course?” Instead of spending hours manually categorizing open-ended comments, you get a usable overview in minutes — leaving more time to actually act on what students told you.

10. Remember What Was Said 🗒️

Jeremy: Use an AI note-taker like Granola to capture transcripts of student meetings, advising sessions, and office hours. Request permission first. You’ll have searchable records of what was discussed, questions that came up, and what you suggested. That’s particularly useful as time passes and it gets harder to remember the nuances of what you talked about.

Lance’s Free Resources for Educators 🎁

Lance is unusually generous in sharing what he’s learned. A few to bookmark:

* AI Syllabi Policies Collection — 200+ real AI policies from faculty across disciplines. See Lance’s post for more context and ideas for applying this.

* Prompts for Educators — a curated tab of tested prompts on his Substack

* Faculty Cohort AI Survey — Lance is gathering data about AI training

* AI + Education = Simplified — his newsletter, worth following for insight

4 More Ideas Worth Noting 💬

1. Nostalgia for the Pre-AI Era 🏺

AI is making polished, professional-looking output trivially easy to produce. That may make what’s authentically human and imperfect more valuable.

“There’s this moment of longing,” Lance said, “for the days when papers students submitted had grammatical errors.” Professors are already nostalgic for flawed student papers.

We may look back on writing of the early 2020s the way we now look at 1900s-era hand-drawn maps or handmade clothing: reminders of a period when you could look at something and know a person made it.

I’ve been thinking about this in terms of pottery and homemade cookies. 🏺🍪Our imperfect things often have more appeal precisely because they’re made with human hands.

“There’s a growing thread about the irrelevance of higher education,” Lance said, “and AI feeds that.” Institutions and creators who figure out how to signal genuine human authorship will have an edge.

Tools are already emerging to signal human authorship. I recently began testing a tool called OKhuman, which verifies that you actually typed something yourself, using your mic to listen to your keystrokes for evidence. This tool’s existence tells you something about where we’re headed.

2. Why Higher Ed is Struggling 🏛️

It’s easy to be impatient with schools that still haven’t developed a coherent AI approach three years in. Lance pushed back on that frustration with useful context. “This happened right on the heels of the pandemic,” he said.

“Every semester it was just like, ‘We have a new format for you.’” Faculty had restructured entire courses repeatedly, switching to remote teaching mid-semester, then hybrid, then back.

Then AI arrived and disrupted assessment design all over again. “If you do that well,” Lance said, of traditional course alignment, “then your course is a deeply intricate web in which everything is related. AI comes in and, in some ways, obliterates the way we do assessments, which means everything else also has to be changed.”

Add funding cuts, political pressure, and leadership distracted by institutional fires, and the picture is complicated.

3. You Can Build Your Own Tools Now 🔧

Lance mentioned that he’s been building custom software, including an MP3 player designed exactly the way he wants it, an RSS reader, and a podcast organizer, even though he doesn’t have a coding background. “This is the first time I’ve really felt like, ‘Oh! I can actually build stuff.”

If you’ve assumed that building custom tools requires a developer, revisit that assumption. AI-powered “vibe coding” tools have lowered the barriers to creating software, making it easier for educators to build what they need.

4. AI May Lead to a New Equity Gap ⚖️

When schools don’t provide AI tools and leave students to their own devices, they inadvertently create a two-tiered system.

As Lance put it: “You have the inequity of the person who’s using the frontier, high-paid model, and the student who gets a limited amount on the free version.” This isn’t an abstract concern. It affects the quality of work students can produce and the skills they develop. For educators and administrators, it’s a reason to push for institutional access, rather than assuming students will figure it out on their own.

Thank you Tom Daccord, Shittu Isaac, Heather Dawn, Robert Hammond, Uyghur Monitor, and many others for tuning into this live video with Lance Eaton, Ph.D.!

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit wondertools.substack.com/subscribe