Students are doing really bad... is it AI?

So I'm 21 years old and my generation has had this very odd coming of life phase of COVID and then straight into AI. And there were signs at first that something was off with the intellectual performance of My generation and the people that have come after us, COVID and post-AI. The last month, some pretty conclusive data came out globally. 15-year-olds all around the world in rich countries showed the worst math and reading scores ever recorded. And in the same month, you have the meta of news blowing up, right? Which is this AI that can write your emails for you and book things in and do useful work basically on your behalf, And I think that these two stories are connected. And in many ways, my generation lost two years of very formative academic time to COVID. And then AI became the bandaid to slap on that wound. And we just never caught up. In fact, we are sliding farther and farther in the wrong direction. And in this video, I want to show you that data. And then I want to show you the counter argument against it. I want to steel man it. And then you can form your own opinion about whether any of this is a problem. That's going to be the video. This is part of our AI Fundamentals series, where we look at the big picture world and evolution. So first, let's hear a word from the sponsor of today's video, and then we'll jump into it. Now, before we keep going, I want to take a moment to tell you about a small biotech that is going after one of the toughest cancers out there. They are the sponsor of today's video, and that is Aptivo Therapeutics, ticker APVO on the NASDAQ exchange. Now, Aptivo focuses on acute myeloid leukemia, or AML, A type of aggressive blood cancer, and a lot of AML patients are older or have pre-existing health conditions, so really intensive chemo is not really an option for them. Now Aptivo's lead drug candidate is called mepletimig, and mepletimig connects cancer cells with your body's own T-cells, so your own immune system can successfully attack the leukemia. And so this is being tested alongside standard treatments. And the latest results here are really encouraging. In TP53, mutated AML, which is one of the hardest forms to treat, 93% of patients saw clinical benefit and 79% reached remission. That's 11 out of 14, and that compares favorably to an existing historical benchmark of about 41%. Then in October, Optivo reported that 10 out of 47 frontline patients, so about 21% advanced to stem cell transplant following the mepletimig combination. And that is a really big deal because transplant can offer the possibility of long-term remission or even a cure. And every one of those patients was initially considered unfit for intensive chemo. So Optivo was planning subject to regulatory feedback to finish dose optimization by the end of this year and then start phase two in the first half of 2027. So there are a lot of big milestones ahead for the company and a growing body of evidence that they're building. Now, mepletimig is still investigational and undergoing trials, and this is an early stage biotech company. That said, you should always do your own diligence to see if an investment like this is a fit for your investment thesis. And you can check out the investor relations page of Optivo here or go to ticker again, APVO on the NASDAQ exchange. That said, thanks to Optivo for their work in fighting AML and for sponsoring this part of the video. And let's get back to it. So that first data point that I mentioned, which is the worst math and reading scores ever recorded came from a test And every two years, this tests 15-year-olds in about 90 countries, typically your wealthier countries globally. And like I said, in September, results came out in among the rich and wealthier OECD countries. Reading fell, I think, 28 points in 10 years. Math fell 22. If you think about that in grade terms, that's like going from an A to a C. Not good. And if we zoom in on the U.S., only one state beat it. Pre-COVID average in math and one different state beat its pre-COVID average in reading. And then in the U.S., 45% of high school seniors did not reach the basic level of math on that test, and their math score was the lowest, again, in recorded history for that test. And it makes sense. The high school seniors right now, they had two years of Zoom during middle school, and then halfway through high school, ChatGPT arrived, right? Lethal combination of being able to get out of doing actual work during a very formative time. And I can say this anecdotally as well, right? I lost my freshman and sophomore years of high school to COVID. And then in college, I had ChatGPT and I used it all the time. It was great. Now, did I use that to not learn? No, but I still did use it. And you can feel the temptation when you're just sitting there and you're like, I could get this. A really hard piece of homework done in 15 minutes by just GPT-ing, right? And so that was my experience, and that was what I saw everyone else my age going through. And then you talk to professors, and all these professors at these schools are saying, hey, something very fundamentally changed in the past five years. Something broke, and it has not recovered. So that's where we're at, and we've identified two causes so far. One is post-COVID AI, the second is COVID itself. I think there's a third, because I don't think COVID actually started any of this. I think COVID was an accelerant to the COVID trend. And as evidence for that, let's look at the test scores of 13-year-olds over a period of 50 years, right? Their scores climbed steadily for decades, and then they peaked about 2012. And something happened around this period where test scores started to come down again. That was eight years before. pandemic. And what happened around 2012? Well, the smartphone started to become an integral of teenage life. My argument there is perhaps COVID was just an accelerant of digitalization and better tools that started with the smartphone. And then maybe your trend was like this. And then with COVID, it became like this. And with ChatGPT, that slope got even steeper. There was this This study out of MIT, where they had kind of Three groups, one group wrote an essay using their brain, one wrote an essay using, I think, Google as support, and then one using ChatGPT. And then they watched their brain activity. And I don't need to tell you what happened. The people who used GPT had zero recall, learned nothing, very little brain activity. And so that's not good. Now, I don't think the distribution is even. And I think this is a really important point. I think, and this is, again, anecdotally in seeing the people around me. You can use AI to learn much better and much faster, but only 10 or 20% of people use AI to accelerate and learn more and learn better and learn faster. So you have this dis-equality, right, in people based on how they use AI, where a lot of them, I think, follow that overarching slope, which is where you get these numbers that are coming. And a small portion of them use it to just accelerate up in a very positive direction. I wanted to focus this video thus far on not one problem of AI as it relates to academic performance, but if we just zoom out, right, there's a lot of other things there. There's the whole socialization problem and having, you know, basically the movie Her if you've seen that. There's the same academic problem then applied to the workplace because now you have all these students going to becoming doctors and lawyers and all these other, you know, important things for society. There's some deeper sociological shift that's happening that I don't think we understand the ramifications of, but is absolutely going to happen. But anyway, I don't want to zoom out too much. I want to keep this video focused on that and that final point there of, hey, actually it can be an accelerant for some people. But before we dive more into that, what's the steel man against this? So first, if test scores started falling around 2012, that has nothing to do with AI. AI did not come out. Until, you know, a decade later. Number two, you need a lot of data over a long period of time to really test these things properly. We don't have that yet. Just has not been enough time. And then number three, people said calculators would kill math, right? People said Google would kill memory. Some of those things kind of came true, but a lot of them are just kind of fear mongering. And I'm, you know, biased in the direction of those things. It's like seeing that happen in so many cases just among. highschoolers and people in, you know, my academic experience. And that's all great. But here's, I think, the final steel man against AI making all of these problems worse, which is just kind of what I said earlier. There was an experiment at Harvard where these students used a custom physics generated writing in an AI chatbot. And it didn't give them answers. It rather prompted them step by step. And it asked them questions. And those students learned about. Twice as much as students in a traditional class, and they did it in less time. And then in Nigeria, there was this six-week AI tutoring pilot. I think that was from the World Bank, and that, on average, raised the English scores of students. And then finally, this one is fun. I mentioned that MIT results that study earlier, which showed, basically, if you use AI, it screws your brain up. But what I didn't tell you is there was another group, and this group was students who first used their own brain to write, and then used AI to revise. And that group showed the strongest brain activity of all. So the tool is not the problem. AI is not necessarily the problem in of itself. And so this finally brings this duality of the problem together, which is a lot of data is coming out. You can form your own opinion, but it's coming out and saying, my generation went through something with smartphones. And on average, they are doing worse academically, and then that will have parallels into their career, again, during a very formative kind of early to mid-academic career and an early actual career. But that's not everyone. That average hides that upper quartile, upper 10% of people who use AI appropriately to become better in their academic careers and their work. And what all of this smartphone AI stuff has done in this, I think, weird summation is it's made thinking optional, it's made focus hyper-optional, and it even is starting to make friendship optional. And if you wrap those things into a bubble, that's not good, right? On average, because of the choice people will make. But those things are optional. You don't have to choose them. You can choose the better path to make all of these things that are an accelerant to you. That, in summary, is what this channel is about. And that speaks to this other very core duality that we've spoken to on this channel before, which is I think the best way to apply AI today is to have Whatever your business, your idea is, your wedge in the market, and then take that and do it better and faster with AI, but it's not to do something that is completely AI from start to scratch, where the idea, the nexus of the business itself was generated by AI, because there's no moat, definitionally. If you could do that with AI, so can everyone else, especially as these tools just continue to emerge. So I think there's a very core parallel here into the career world, even if you're not a student, about how AI either thinks for you or it helps you think and do. As far as this all relates to the AI Fundamentals series, I think this was a little bit more of a personal one than usual. That covers where we're at, I think, in the trend over the past 15 or so years. If we look toward the future, you are going to see those two forces compete against one another. For example, you're going to see AI tutors in third world countries where there are just not enough teachers and their scores are going to go up. And among wealthy countries, you are going to see I think this continued just kind of downward, very aggressive slide in performance. But also the school systems will start fighting back. They're going to start just basically throwing out technology and making everything, making everyone do everything in person with a pencil like the old days. So there is some head-butting that's going to go on here to, I think, try and regulate this effect. I don't know if that will work, but it is a very important dynamic. for society and for a world that is going to play out in a very big way over the next 5, 10, or 20 years. I leave you on that idea. I hope you find this food for thought, and I appreciate you being here.