Future of math tutoring, especially considering AI

General Forum
1

I’m on disability in the U.S. They allow me to work a little bit without losing my benefits. 6 years ago I discovered that I was able to do math and CS tutoring and it was great for my mood to have some responsibilities and contribute to kids’ lives.

I started with math tutoring, mainly prealgebra and algebra, but honestly, I didn’t do it well as I hadn’t studied any pedagogy yet. I switched to doing CS: competitive programming and AP Computer Science. I worked as a programmer before becoming disabled, and I studied pedagogy at this point, so I am pretty good at CS tutoring.

However, I’m burning out on CS tutoring and I’m thinking of switching back to math.

  • CS is all online. In-person tutoring has a greater effect on my mood, and I just enjoy it more. When I last did math tutoring in 2019, there was a demand for in-person math tutoring.
  • Competitive programming has been getting rapidly harder in the past few years as more kids take it up and they need hard tests to filter out the best. What that means is there are a lot of kids who come to me, but they lack the raw talent and available practice time to score well on today’s tests, so they and their parents are frustrated.
  • Most of my AP Computer Science students expect me to do the work for them.
  • Tutoring competitive programming takes a lot more unpaid preparation time for various reasons.

So, can I be a good math tutor? I attended a strong engineering college and took a ton of advanced math, so I’m pretty good at math.

However, I haven’t used math in 30 years (except for tutoring algebra in 2018-2019). Right now I’m reviewing precalculus and studying precalculus pedagogy so I can add that class to my offerings. Next I hope to review calculus and calculus pedagogy.

I’m still doing CS tutoring for now and plan to update my Wyzant profile to indicate I’m doing math too, starting this fall.

However, as I consider the switch back to math, I’m concerned about AI taking over. I’m sure that some parents will love the affordability of AI. On the other hand, some people will argue that a good in-person human tutor can add that human touch. Maybe the human touch will still be of great benefit even ten years from now. Or at least, the affluent families will believe it’s still of benefit and will hire human tutors.

I do live in an affluent area.

By the way, I only have 6 to 8 lessons a week due to my difficulties working. I’m not trying to make a living at this. The supplemental money is very helpful for paying my medical expenses, and having work responsibilities is very good for my emotional outlook (it SUCKED when I wasn’t working at all).

So I’m curious what people here think about the viability of in-person math tutoring.

6 Likes
2

AI is likely to be better at programming than math; a lot more resources are out into the former than the latter

1 Like
3

The other thing is that with programming when it makes a syntax error the compiler will typically find and announce it and you can feed it right back to the AI and tell it to fix it. Any programmer who knows what they are doing is still going to test the code for logic errors as well and tell the ai to fix it.

With math a student could compare a final answer from somewhere like wolfram to the answer and explanation that the ai gives, then bounce it back to the ai if the answer is different, but they would have to think to do that.

4

I feel there should be no need to worry unless AI can create good Youtube videos. In general, as a College math instructor, I see evidence that students needing tutors stopped learning math by reading anything a while back.

6

I have been thinking about this a lot, and thought about doing a startup in that area. I think there’s a lot of enthusiasm in tech for GPT-4 to just swoop in and solve this, and there are startups like Byte Learn offering chat clients. But there are a lot of good components to tutoring that I don’t see anyone working on. Especially for younger students, chat text won’t really help much. Good tutors point at parts of your work and help you see which parts you did right, and where it went wrong. They can write on your page and talk. They know their students and have some concept of what knowledge content they know and what they don’t know yet. In data science circles, we call that task “knowledge tracing”, and the current state of the art is really rudimentary, and hasn’t really substantially improved in the last decade, even with all the other advances in AI. It’s really a totally separate capability, that I don’t see anyone working on at the moment. There are a lot of other capabilities that would be needed that I don’t see anyone talking about.

I think there’s a lot of potential, but that doesn’t mean it’ll happen any time soon or even that people are working on it. Personally I think that would be great for the world. We know math is really important and that 1-on-1 tutoring is immensely effective for improving achievement, but also prohibitively expensive at scale. I think there will always be demand for human tutors, even if AI gets good.

1 Like
8

Yeah don’t worry, A.I. is really shitty at maths tutoring. By the time I finish re-prompt GPT-4 for the nth time to finally grasp what it’s trying to say, a person could’ve helped me understand a maths concept three times back-to-back.