While interesting and perhaps aspirational, this article exists in a different universe than the vast majority of K-12 education in the US. "When students experience 2 to 5x acceleration in K-12" is like saying "when students learn unicycling and sword-swallowing in K-12." In my school district as in many others the trend has been to deceleration: we've gotten rid of advanced math classes so all kids are in the same class taught the same things; algebra is increasingly pushed back a year from what it used to be. My kid's school's standard 11th grade literature curriculum does not assign whole novels. (Thankfully, he's in the IB program, and they actually read books. The teacher who told me this added that the actual teachers in the school hate not assigning novels, and of course this hurts students.) And so on -- outside the fever dreams of silicon valley, there isn't a push to accelerate, or even to allow students to maximize their potential. This is what must be countered, or at least acknowledged.
As a private tutor for 40 years, I totally agree with this statement that "outside the fever dreams of silicon valley, there isn't a push to accelerate, or even to allow students to maximize their potential."
I've often wanted to say this, and I appreciate you doing so. All one needs to do is compare curriculum from the 1980s to today's curriculum to see this decline, and how little is required of students today.
I believe that students become bored, as nothing tests their ability to learn, so they developed an attitude of who cares about school, or even learning, or so I have found with my clients. I've been encouraging those who still have interest to access the freely available courses, and learn at their own pace in what interests .These students will have expectations of accelerated learning in post secondary.
I'd like to add that, as a result of multiple avenues for learning that are available now, home educated students will definitely have even higher expectations.
A lot of the structural problems you describe seem to be linked specifically to how US higher education works. When I studied in Germany 30 years ago (no idea how it works toady), every class was pass/fail, and I could just walk into the final of a class without ever having registered for it or attended, take the exam, and if I passed I got credit for the class. I got credit for many of my classes in this way, and I shaved over a year off my graduation time. It didn't create any problems for anybody.
(Also, interesting from a grade inflation perspective, many of the harder classes had 30-50% failure rates. The exams were not easy by any stretch of the imagination. But you could study for them on your own time and then just take them for credit.)
I did BA + MA in germany from 2018 - 2025 (coincidentally doing my PhD here now in higher education research for economics). Lecture based classes are still like this. Professors just try to figure otu how to incentivize people going to class, like not uploading their slides, etc. Seminars still often require attendance and usually paper/project.
Yeah, every system has problems. The high failure rates in the classes I took were in part due to students not attending class, not doing the optional weekly homeworks, and then not knowing anything when taking the exam. But they didn't worry too much about it because they knew they could just retake the exam without any penalty. And so some students took years to finish their degree.
Fascinating questions. I see significant value in paring down the traditional undergraduate degree to 2 years for most students -- those who want to get a "normal" job and are going to college because it's the default for their social class. For specialists, however, these 2 years can be supplemented with 2 years of research at what we currently consider a master's level. China already does something like this with, e.g., med students, whose undergrad includes specialized medical education.
I'm also thinking about my own undergrad, which was a pretty unusual Great Books curriculum (https://www.sjc.edu/academic-programs/undergraduate/great-books-reading-list) that would be almost impossible to compress into 2 years. At the same time as most programs benefit from shrinkage (since most people go to college for the piece of paper and the connections), the few students who are genuinely passionate about their area of study may choose to stick with a 4 year option that signals deep engagement, signal-boosted compared to the new 2-year norm.
Yes, and agreed that there are degrees not worth racing through, and SJC is one! (I was there this summer and wish I could spend more time reading deeply.) But for other people, there is no fit. The 4-year degree is a Procrustean bed!
I was thinking something like this. My undergraduate courses at Hopkins had a lot of reading (humanities and social sciences). It would have been difficult to compress them.
These are really tough questions. If you haven't listened to this lengthy podcast on Alpha School, it's illuminating. I still don't quite understand how they teach writing and research so I don't know if the 2x plus rates apply across the board, but it's a really interesting listen nonetheless.
So many students today take many more AP and college level classes in high school that they enter college as sophomores or even juniors. Are these students graduating in 2-3 years? I don’t think there is a significant portion that do. I also wonder about networking and exposure to the classroom of ideas the campus presents. Seems to me change won’t come so fast, or be so intense.
It sounds like the prime issue for encouraging reform/education of higher education(and other) education is the accreditation process. The accreditors are simply protecting the entrenched model. For example, the new University of Austin appears to be having trouble getting accredited, but does anyone seriously believe that the education it provides is inferior to the thousands of random accredited colleges? Also, I see better lectures on YouTube on a variety of topics than I received at some of the most elite universities. We need to move to a system with a wide variety of ways that students can prove their competency in various fields. Obviously, this will vary from field to field. I think employers are already discounting the information they receive from seeing a degree on a resume, but they would like to see meaningful credentials.
These are far-sighted questions and the whole problem is nicely thought through. I'm struck, though, that we're not sure whether AI will leave us with super-students (as hypothesized here) or students who are super stupid (which seems to be what everyone else in my Substack feed is worried about). I'll definitely be interested in whether my colleagues in mathematics and the natural sciences start seeing super-advanced students in their classes, or a flood of students testing out of requirements.
The likelihood as I see it is being super advanced in some fields and super behind in others. I once heard these kids called "lopsided" though I don't like the term. Our educational system is not prepared for them beyond requiring sub-standard general education courses for big $.
Amazing post - thank you for the thought you put into this. The real estate / infrastructural value of college seems extremely valuable even in an AI-first world so I’m confident we’ll find a use for it. Given that and other economic factors, I doubt we’ll move towards a smaller student body for most major universities (maybe some smaller ones will). Hopefully AI-first learning will allow more students to free themselves of the obnoxious expectation to attend class and instead they will learn on their own and simply pass the test. That’s how I did college and it enabled me to spend most of my weekday time working.
We could be extreme and ask, if students in higher education are able to access and accelerate gaining the knowledge offered in the current curriculum, why would anyone bother attending higher education at all, when the LLMs can tutor you at home, at your level?
How about this; Why not offer more within a course than what is offered now?This is an opportunity to change a system that was not optimal anyways.
Every course has an outline curriculum, with readings and assignments, and a structured lecture. Now, let's retain the four years to attain a degree, maintain the required material to learn, but students will gain proficiency in the knowledge twice as fast.
I am a private tutor. My clients often state that higher education is immensely frustrating. By the time the course material forms into a cohesive whole, the course is done. Do the exam, rinse and repeat. Clearly, upper level courses require a firm grasp of lower level. Students want to engage with the material right now, when they've just learned it. But, there is no time and this lack of time to integrate knowledge is frequently raised in after class pub discussions. Let’s have those discussions in class with the prof, who knows more than the students.
So, as a teacher, have you ever wished you had more time for the courses you teach, to add something to it, within that same required time frame? Perhaps, you'd like each student to research one area, and do a lecture to the class? Would you like to arrange for a debate on one core area? Would you integrate a class with another professor, or researcher, say even in a different faculty, to do some of the inter-learning that is so often suggested as valuable, but rarely happens? To me, there is a potential here for a broader expansion of knowledge acquisition. Higher education is knowledge acquisition that scaffolds upon previous learning. In every field, research compiles almost daily, yet it's challenging to add that in and still cover the fundamentals for a Bachelor's degree. There is a wealth of knowledge available but one only skims it to attain a degree. Fundamentals learned quickly means time is left for alternative and iterative ways to learn.
Let's not limit reimagining higher learning by looking at the current category boxes, and trying to fill those; let's create new bones, irregular sized boxes, maybe even a few circles and triangles.
AI is changing the future of work, and via that the nature of how we educate. We can change the 'boxes' to incorporate our human qualities, like curiosity, and focus less on productivity, thereby creating a richer learning experience filled with meaning. This is especially valuable when we consider that, in a world where AI dominates, human purpose, unearthing what matters, matters even more.
(Meant to be a reply to Ethan‘s second comment below) I would note that in the second and third paragraphs you can replace the word AI with the word teacher and it applies about the same for the average teacher and below. There’s still no sign of AI development slowing down, and we should expect the opposite given how much money their companies are spending on it, so we have to take seriously the idea that in one or two years the AI might actually be better than the average teacher, if it’s not already. Also most kids aren’t using the cutting edge model but rather the free version.
Great framing Hollis. I wonder though if we have to zoom out even further & first ask what business & society of the future need from higher ed before we transform ed? Our current ed syatems & institutions were designed/optimized for the industrial era with some ptaches as the world changed. Maybe its not even a 2 min mile, perhaps its pickleball that we need to prepare for?
It's a good question, though I take the position that higher education spends too much time "catching up" to industry and if we started preparing for pickleball something new will have taken its place by the time students graduate. No, let industry do its own job training (https://hollisrobbinsanecdotal.substack.com/p/remember-corporate-training-programs) Higher ed needs to educate.
Your deep dive, long view, marathon race to foresee outcomes of “how does a university assess whether students claiming accelerated mastery actually possess competent knowledge versus credentials from variable-quality programs” is heartening to read. To witness your caring about student outcomes by questioning the process is professionally noble. Grateful for your dedication, Hollis.
I doubt we're going to see the Alpha School model work at scale. As the SlateStarCodex review of Alpha School noted, it doesn't work that well outside of a small group of carefully self-selected gifted students.
AI in education is going to go the way of the MOOCs: a small percentage of students, who will get lauded and held up as exemplars, will use AI to learn a great deal and enhance their abilities.
The vast majority of students will struggle, learn less, and fall further behind. But first, there will be much disruption as campuses rush to "do AI" in haphazard fashion.
Employers are already working to suss-out AI-generated resumes and interview responses. The students who are using AI judiciously and actually learning the content are going to have a much easier time standing out in practice, if not in school (since universities are terrified of lawsuits and will refuse to hold students accountable).
A thought-provoking article. And here's a thought in reply: how many more students than at present will be capable of accelerated learning? If more top students can finish college by age 20, fine, but I don't foresee a general acceleration of learning such that median entering or exiting college students a decade hence will be perceptibly better educated than their counterparts today.
While interesting and perhaps aspirational, this article exists in a different universe than the vast majority of K-12 education in the US. "When students experience 2 to 5x acceleration in K-12" is like saying "when students learn unicycling and sword-swallowing in K-12." In my school district as in many others the trend has been to deceleration: we've gotten rid of advanced math classes so all kids are in the same class taught the same things; algebra is increasingly pushed back a year from what it used to be. My kid's school's standard 11th grade literature curriculum does not assign whole novels. (Thankfully, he's in the IB program, and they actually read books. The teacher who told me this added that the actual teachers in the school hate not assigning novels, and of course this hurts students.) And so on -- outside the fever dreams of silicon valley, there isn't a push to accelerate, or even to allow students to maximize their potential. This is what must be countered, or at least acknowledged.
This is the conversation I hope to provoke yes!
To be clear, I think it's great that you're working on this subject and encouraging discussions of these questions.
As a private tutor for 40 years, I totally agree with this statement that "outside the fever dreams of silicon valley, there isn't a push to accelerate, or even to allow students to maximize their potential."
I've often wanted to say this, and I appreciate you doing so. All one needs to do is compare curriculum from the 1980s to today's curriculum to see this decline, and how little is required of students today.
I believe that students become bored, as nothing tests their ability to learn, so they developed an attitude of who cares about school, or even learning, or so I have found with my clients. I've been encouraging those who still have interest to access the freely available courses, and learn at their own pace in what interests .These students will have expectations of accelerated learning in post secondary.
I'd like to add that, as a result of multiple avenues for learning that are available now, home educated students will definitely have even higher expectations.
A lot of the structural problems you describe seem to be linked specifically to how US higher education works. When I studied in Germany 30 years ago (no idea how it works toady), every class was pass/fail, and I could just walk into the final of a class without ever having registered for it or attended, take the exam, and if I passed I got credit for the class. I got credit for many of my classes in this way, and I shaved over a year off my graduation time. It didn't create any problems for anybody.
(Also, interesting from a grade inflation perspective, many of the harder classes had 30-50% failure rates. The exams were not easy by any stretch of the imagination. But you could study for them on your own time and then just take them for credit.)
I did BA + MA in germany from 2018 - 2025 (coincidentally doing my PhD here now in higher education research for economics). Lecture based classes are still like this. Professors just try to figure otu how to incentivize people going to class, like not uploading their slides, etc. Seminars still often require attendance and usually paper/project.
Yeah, every system has problems. The high failure rates in the classes I took were in part due to students not attending class, not doing the optional weekly homeworks, and then not knowing anything when taking the exam. But they didn't worry too much about it because they knew they could just retake the exam without any penalty. And so some students took years to finish their degree.
Fascinating questions. I see significant value in paring down the traditional undergraduate degree to 2 years for most students -- those who want to get a "normal" job and are going to college because it's the default for their social class. For specialists, however, these 2 years can be supplemented with 2 years of research at what we currently consider a master's level. China already does something like this with, e.g., med students, whose undergrad includes specialized medical education.
I'm also thinking about my own undergrad, which was a pretty unusual Great Books curriculum (https://www.sjc.edu/academic-programs/undergraduate/great-books-reading-list) that would be almost impossible to compress into 2 years. At the same time as most programs benefit from shrinkage (since most people go to college for the piece of paper and the connections), the few students who are genuinely passionate about their area of study may choose to stick with a 4 year option that signals deep engagement, signal-boosted compared to the new 2-year norm.
Yes, and agreed that there are degrees not worth racing through, and SJC is one! (I was there this summer and wish I could spend more time reading deeply.) But for other people, there is no fit. The 4-year degree is a Procrustean bed!
I was thinking something like this. My undergraduate courses at Hopkins had a lot of reading (humanities and social sciences). It would have been difficult to compress them.
These are really tough questions. If you haven't listened to this lengthy podcast on Alpha School, it's illuminating. I still don't quite understand how they teach writing and research so I don't know if the 2x plus rates apply across the board, but it's a really interesting listen nonetheless.
https://podcasts.apple.com/us/podcast/invest-like-the-best-with-patrick-oshaughnessy/id1154105909?i=1000723564395
Thank you this is interesting!
So many students today take many more AP and college level classes in high school that they enter college as sophomores or even juniors. Are these students graduating in 2-3 years? I don’t think there is a significant portion that do. I also wonder about networking and exposure to the classroom of ideas the campus presents. Seems to me change won’t come so fast, or be so intense.
It sounds like the prime issue for encouraging reform/education of higher education(and other) education is the accreditation process. The accreditors are simply protecting the entrenched model. For example, the new University of Austin appears to be having trouble getting accredited, but does anyone seriously believe that the education it provides is inferior to the thousands of random accredited colleges? Also, I see better lectures on YouTube on a variety of topics than I received at some of the most elite universities. We need to move to a system with a wide variety of ways that students can prove their competency in various fields. Obviously, this will vary from field to field. I think employers are already discounting the information they receive from seeing a degree on a resume, but they would like to see meaningful credentials.
These are far-sighted questions and the whole problem is nicely thought through. I'm struck, though, that we're not sure whether AI will leave us with super-students (as hypothesized here) or students who are super stupid (which seems to be what everyone else in my Substack feed is worried about). I'll definitely be interested in whether my colleagues in mathematics and the natural sciences start seeing super-advanced students in their classes, or a flood of students testing out of requirements.
The likelihood as I see it is being super advanced in some fields and super behind in others. I once heard these kids called "lopsided" though I don't like the term. Our educational system is not prepared for them beyond requiring sub-standard general education courses for big $.
Amazing post - thank you for the thought you put into this. The real estate / infrastructural value of college seems extremely valuable even in an AI-first world so I’m confident we’ll find a use for it. Given that and other economic factors, I doubt we’ll move towards a smaller student body for most major universities (maybe some smaller ones will). Hopefully AI-first learning will allow more students to free themselves of the obnoxious expectation to attend class and instead they will learn on their own and simply pass the test. That’s how I did college and it enabled me to spend most of my weekday time working.
I wrote a book where I use the presence of clock towers on campus to illustrate the history of time-based proxies for learning. Great post.
We could be extreme and ask, if students in higher education are able to access and accelerate gaining the knowledge offered in the current curriculum, why would anyone bother attending higher education at all, when the LLMs can tutor you at home, at your level?
How about this; Why not offer more within a course than what is offered now?This is an opportunity to change a system that was not optimal anyways.
Every course has an outline curriculum, with readings and assignments, and a structured lecture. Now, let's retain the four years to attain a degree, maintain the required material to learn, but students will gain proficiency in the knowledge twice as fast.
I am a private tutor. My clients often state that higher education is immensely frustrating. By the time the course material forms into a cohesive whole, the course is done. Do the exam, rinse and repeat. Clearly, upper level courses require a firm grasp of lower level. Students want to engage with the material right now, when they've just learned it. But, there is no time and this lack of time to integrate knowledge is frequently raised in after class pub discussions. Let’s have those discussions in class with the prof, who knows more than the students.
So, as a teacher, have you ever wished you had more time for the courses you teach, to add something to it, within that same required time frame? Perhaps, you'd like each student to research one area, and do a lecture to the class? Would you like to arrange for a debate on one core area? Would you integrate a class with another professor, or researcher, say even in a different faculty, to do some of the inter-learning that is so often suggested as valuable, but rarely happens? To me, there is a potential here for a broader expansion of knowledge acquisition. Higher education is knowledge acquisition that scaffolds upon previous learning. In every field, research compiles almost daily, yet it's challenging to add that in and still cover the fundamentals for a Bachelor's degree. There is a wealth of knowledge available but one only skims it to attain a degree. Fundamentals learned quickly means time is left for alternative and iterative ways to learn.
Let's not limit reimagining higher learning by looking at the current category boxes, and trying to fill those; let's create new bones, irregular sized boxes, maybe even a few circles and triangles.
AI is changing the future of work, and via that the nature of how we educate. We can change the 'boxes' to incorporate our human qualities, like curiosity, and focus less on productivity, thereby creating a richer learning experience filled with meaning. This is especially valuable when we consider that, in a world where AI dominates, human purpose, unearthing what matters, matters even more.
This is the conversation we should be having!
(Meant to be a reply to Ethan‘s second comment below) I would note that in the second and third paragraphs you can replace the word AI with the word teacher and it applies about the same for the average teacher and below. There’s still no sign of AI development slowing down, and we should expect the opposite given how much money their companies are spending on it, so we have to take seriously the idea that in one or two years the AI might actually be better than the average teacher, if it’s not already. Also most kids aren’t using the cutting edge model but rather the free version.
Great framing Hollis. I wonder though if we have to zoom out even further & first ask what business & society of the future need from higher ed before we transform ed? Our current ed syatems & institutions were designed/optimized for the industrial era with some ptaches as the world changed. Maybe its not even a 2 min mile, perhaps its pickleball that we need to prepare for?
It's a good question, though I take the position that higher education spends too much time "catching up" to industry and if we started preparing for pickleball something new will have taken its place by the time students graduate. No, let industry do its own job training (https://hollisrobbinsanecdotal.substack.com/p/remember-corporate-training-programs) Higher ed needs to educate.
Your deep dive, long view, marathon race to foresee outcomes of “how does a university assess whether students claiming accelerated mastery actually possess competent knowledge versus credentials from variable-quality programs” is heartening to read. To witness your caring about student outcomes by questioning the process is professionally noble. Grateful for your dedication, Hollis.
I doubt we're going to see the Alpha School model work at scale. As the SlateStarCodex review of Alpha School noted, it doesn't work that well outside of a small group of carefully self-selected gifted students.
AI in education is going to go the way of the MOOCs: a small percentage of students, who will get lauded and held up as exemplars, will use AI to learn a great deal and enhance their abilities.
The vast majority of students will struggle, learn less, and fall further behind. But first, there will be much disruption as campuses rush to "do AI" in haphazard fashion.
Employers are already working to suss-out AI-generated resumes and interview responses. The students who are using AI judiciously and actually learning the content are going to have a much easier time standing out in practice, if not in school (since universities are terrified of lawsuits and will refuse to hold students accountable).
Should we also be considering what to do if a student shows up who is incapable of reading something longer than one page? Serious question.
So many other writers are addressing that question!
A thought-provoking article. And here's a thought in reply: how many more students than at present will be capable of accelerated learning? If more top students can finish college by age 20, fine, but I don't foresee a general acceleration of learning such that median entering or exiting college students a decade hence will be perceptibly better educated than their counterparts today.