Last Tuesday I sat through a 75-minute organic chemistry lecture where my professor read directly from slides he made in 2019. Slides I already had access to. Slides I could've read in 12 minutes.
75 minutes of my life. Gone.
And look, I get it. Not every lecture is bad. I've had professors who genuinely changed how I think about a subject. But let's be honest about the ratio here. For every one professor who makes you lean forward in your seat, there are like four who make you wonder why you're paying $400 per credit hour to watch someone read a PowerPoint.
The Daily Pennsylvanian published a piece this week that said something I've been thinking for months: if a chatbot can outperform your lecture, the problem isn't AI. The problem is the lecture.
Damn right.
The Lecture Model Is Broken (And Has Been for a While)
Here's what a typical lecture looks like in 2026:
- Professor talks for 50-75 minutes
- You try to take notes while also understanding what's being said (pick one, you can't do both well)
- You zone out for 10 minutes somewhere in the middle and miss a concept that everything else builds on
- You leave confused but too tired to go to office hours
- You teach yourself the material from YouTube at 11pm anyway
Sound familiar? Yeah.
The research backs this up too. Students retain about 10-20% of what they hear in a traditional lecture after 48 hours. That's not a study method. That's a memory leak.
Meanwhile Harvard just published a whole thing about how AI shortcuts are threatening cognitive development. And they're right, partially. If you're just copy-pasting your homework into ChatGPT and submitting whatever comes out, you're not learning anything. You're just laundering your GPA.
But there's a massive difference between using AI to skip learning and using AI to replace a bad delivery system.
What AI Actually Does Better Than a Lecture Hall
I started using AI study tools about 8 months ago. Not to cheat. To actually understand my textbooks. Here's what changed.
It goes at my speed. My neuroscience professor covers 3 chapters per lecture. I need about 20 minutes on chapter 7 and about 2 hours on chapter 8. A lecture treats both the same. AI doesn't.
I can ask dumb questions. You ever sit in a 200 person lecture hall and think "wait what does that mean" but not raise your hand because the professor already explained it and you don't want to be that person? With AI there's no judgment. I've asked the same concept 4 different ways until it clicked. Try doing that in a lecture.
It actually checks if I understood. After I read a chapter on textbooks.ai, it generates practice questions based on what I just covered. Not generic textbook questions from the back of the chapter. Questions about the specific stuff I was struggling with. A lecture just... ends. And hopes you got it.
It connects things across chapters. Last week I was studying for a psych exam and asked how classical conditioning relates to the cognitive development stuff from 3 weeks ago. Got a clear answer in 30 seconds. In a lecture, those connections almost never get made because each class is its own island.
The Lectures Worth Showing Up For
I'm not saying burn down every lecture hall. Some classes genuinely need a human in the room.
Lab sciences where you're doing hands-on work. Seminars where the discussion IS the learning. Classes with professors who challenge your thinking in real time and adapt based on the room. Those are irreplaceable.
My philosophy professor does this thing where she'll present an argument, wait for someone to disagree, then systematically destroy both sides until we're all confused in a productive way. No AI does that. Not yet anyway.
But the 300-person intro to biology lecture where someone reads from a textbook? Come on. We all know that could be an email. Or better, it could be an AI-powered study session where you actually engage with the material.
How to Use AI Without Becoming Stupid
Harvard's concern about AI making students dumber is valid if you use it wrong. Here's the line I draw.
Use AI to understand, not to produce. Read the chapter. Struggle with it a little. Then use AI to fill the gaps. Don't start with AI and skip the struggle entirely. The struggle is where learning happens.
Test yourself constantly. After any AI study session, close everything and try to explain the concept out loud. If you can't, you didn't learn it. You just watched AI learn it. Tools like textbooks.ai build this in with auto-generated quizzes, which honestly saves me from lying to myself about how much I actually know.
Don't outsource your thinking. Use AI to understand concepts faster. Don't use it to write your essays or solve your problem sets. The point of those assignments is to make your brain do the work. Skipping that is like paying someone to go to the gym for you.
Mix your sources. AI plus your textbook plus lecture notes plus study groups. No single source is enough. The students who are thriving right now are the ones who use AI as one tool in a bigger system, not their entire system.
The Real Problem Nobody Wants to Talk About
Colleges charge $30,000-$80,000 a year. A huge chunk of that supposedly pays for instruction. If AI can deliver better instruction for the foundational stuff, what exactly are we paying for?
That's the question that makes administrators nervous. And honestly, it should.
I'm not saying college is worthless. The connections, the experiences, the credentials, those matter. But the actual teaching? For a lot of classes, we've been settling for mediocre delivery because there was no alternative.
Now there is one.
The schools that figure this out first, the ones that let AI handle the content delivery and free up professors to do the stuff humans are actually good at (mentoring, discussion, research), those are going to win. The ones that pretend nothing has changed are going to keep charging $400 per credit hour for someone to read a PowerPoint.
What I Actually Do Now
My system for this semester:
- Before lecture, I read the chapter on textbooks.ai and take the practice quiz. Takes about 40 minutes per chapter.
- I go to lecture with questions already formed. Instead of trying to absorb everything, I'm listening for the stuff I didn't get from my pre-read.
- After lecture, I do a 15-minute review session with AI focusing on whatever confused me.
- Before exams, I use spaced repetition on the concepts I flagged as tricky.
My GPA went from a 3.1 to a 3.6 in one semester. Not because AI did my work. Because AI fixed the broken parts of how I was being taught.
That's not cheating. That's adapting.
The Bottom Line
Bad lectures were never sacred. They were just all we had. Now we have options. The students and schools who embrace that are going to come out ahead. The ones who cling to "but this is how we've always done it" are going to keep hemorrhaging students who figured out they can learn more in 40 minutes with AI than in 75 minutes in a lecture hall.
Your move, higher ed.