I review dozens of EdTech products every week and almost none move the needle like Sizzle. It is simple to use. You take a picture of a STEM problem and it solves it for you. The most impressive aspect is that it does so in a step-by-step process (see picture below). This is not what I’m used to with AI, where you get an answer and take it or leave it.
For other EdTech companies, the question is how to add automated STEM assistance to your product. It will have to be integrated so that teachers can see a holistic view of their students’ work. How much of the solution did the student do versus the automated tutor? How many hints did each student need? Where in the solution process were the most hints requested? Does it make more sense to compare images (student work vs solution) to find the relevant solution? If you are doing image searches, does it make sense to suppress the numbers to find a generic problem corpus? Or, does it make more sense to convert the images into text and LaTeX? I need to do more research in this area. Please reach out if you have done this type of work.
The generative nature of the latest form of AI could be exciting for education. Imagine a student is working towards developing proficiency in a STEM skill. They reach mastery through a combination of teacher-provided and automated assistance. The next step could be to re-enforce mastery by becoming a teacher to other students (peers or younger). Students use AI tools to generate problem sets for other students. And, they work out solutions.
