As I've been looking around for practical use cases of Generative AI that can impact people today, I find myself getting more and more convinced that Education is going to be disrupted by GenAI. A case in point that this is already happening is Chegg. A lot has been written about this.
In July 2023, the company’s CEO admitted at a Fortune conference that Chegg had seemingly become a “poster child for getting your ass kicked in the public markets by AI.”
This was an existential crisis for Chegg and they reacted quickly.
“I called our team and said, ‘We’ve got to get together. We’re literally going to rip our product down to its roots and we’re going to reinvent it on the fly,’” - Dan Rosensweig, CEO of Chegg.
That being said the stock has not been doing great. It has dropped 75% since last November and 90% since ChatGPT came out.
My younger son is a junior in high school and we are getting to the time where we needed to sign him up for a tutor to do some SAT test prep. This is what we had done with our older son but I was on the fence as I was not sure the tutor had helped my older son. The tutor came highly recommended and he was good. My son put in all the hard work but I always felt there was something about the way he was teaching that was not right for my son.
As all this was going on, I asked myself the same question I ask myself for most such things nowadays - How can GenAI help? This got me doing some research and sure enough I came across Acely.ai an app that helps with SAT test prep (and more) and has GenAI infused in it. It was created by some ex. Stanford graduates and I was curious how it would work.
The one thing about any application of GenAI is that you have to be willing to experiment. There is no guarantee it will work and this is all so new that there isn't enough data out there on success or failure. So people leveraging it are on the bleeding edge and willing to make the time and money investment, knowing that it might not work.
I asked my son to go to the website and see what he thought and if he saw value in signing up and using it and he did. It asked him to enter his PSAT scores, his goals and it came up with a plan to be ready for the SATs this weekend.
Let's see how he does as that will be the real test but here are a few positives already -
My son has spent, on average, a couple hours each day on Acely. This in itself is a huge win. It shows that the platform is engaging and makes test prep fun. I remember buying him one of those Princeton Review books and he hardly touched it. Something about paper etc. did not work for him.
It has gamified test prep. He has a baseline score and as he answers questions he can see himself improving.
When he gets a wrong answer or even if he gets it right but is not sure, he can ask it questions and get clarifications. He can go back and forth like we do with ChatGPT. This is the biggest impact of GenAI. He could be answering a question at midnight and need a clarification on a topic and he can get it instantly. No tutor can provide this level of service.
It is able to pinpoint his areas of weakness and give him a ton of questions on just that area. This focused learning is key as this will make a big impact on his score. Plus he is learning the underlying concepts on topics that he was not sure about.
It is able to give him harder questions as he gets better and also allow him to ask for a harder (or easier) version of the same question.
You cannot beat the price point. It would cost thousand(s) to get a tutor and this was $299 for 6 months.
This got me thinking on why this application of GenAI seems to work so well and here are a few things that I thought about -
The knowledge domain is fixed. It is about SAT test prep. There is a ton of information on past tests. The categories are fixed and well understood and it is a lot of unstructured data.
The alternative is a human that is expensive, is available for a limited time and on their schedule.
The human may or may not be able to teach you the way you like to learn. There is a variability in the quality of the human. If we apply a bell curve, if you get lucky you might get that tutor who is in the top 20% but chances are you can get someone who is in the bottom 80%.
While digital test prep is not a new thing, adding GenAI to it solves the one key missing piece - have something you can go back and forth with when you are stuck, so you can learn the underlying concepts. In the past, you could take a test, it could establish a baseline, it could give you questions in your areas of weakness - with AI it can dynamically tailor questions on the fly and it can help you when you get stuck.
Coming back to the broader topic of GenAI in Education - Acely is now rolling out test prep for AP courses. I was talking to a friend who has invested in a company that is using GenAI to generate lesson plans for schools.
Why I feel GenAI will disrupt education (this will take a lot of time as it will have a huge impact on teaching jobs, there will be political input to this etc.) -
Lesson plans are standardized.
Quality of the education is dependent on the teacher.
It's very expensive.
Kids are used to learning digitally.
Curious if people have an opinion on this topic.