Einstein is here.
Einstein is an AI bot described this way: “Einstein is a computerized AI. He logs into Canvas every day, watches lectures, reads articles, writes papers, participates in discussions, and submits your homework—automatically.”
Yes, the product claims that you can complete any task in the Canvas course itself. There may be exceptions—such as uploading your video—but removing such an obstacle can ultimately be well-solved.
What does this mean? I think it means less and more than some people might think.
I’m headed to a speaking gig, which leaves me short on time to develop a full, thoughtful post, but let me throw out a series of thoughts to perhaps contextualize what I think is going on here.
- AI agents that can log into an LMS and complete tasks are nothing new. Marc Watkins and I talked about them in a Q&A last November, referring to Anna Mills’ earlier work.
- We don’t know exactly what this is. Companion.AI, which appears to be an Einstein-hosted business, is the open project of Advait Paliwal, a 2024 graduate of Michigan State.
- Whatever it is, it doesn’t seem like a big, business-backed project. Indeed, given the nature of automated coding tools and the easy integration of LLMs into the system, it is easily believable that one person or a small group could investigate something that looks like this and does it transparently.
- I haven’t tried it yet because I don’t have access to a safe Canvas tutorial shell, but it should be something. We should know how skilled these agents are.
- Yes, writing about it, or posting about it on BlueSky (as I and others in my circle have been doing) is doing a measure of product marketing, but these things—especially if they work—will eventually spread anyway.
- There is no winning the arms race against automated AI technology. Not only will acquisition fail, but investing significant energy in acquisition is a waste of time and resources.
- Analogous evaluation is only a partial defense against these agents, and choosing to fall back exclusively on these methods risks limiting the kind of learning that students experience.
- The deep roots of AI’s existence for an Einstein-like agent is the commercial model of education that reduces the school to authenticity. Einstein promises to do “busy work so you don’t do it.” That students see an important part of what happens at school as busy work is a problem that needs to be addressed.
- It is obvious to those who use a lot of seasoning that studying nothing while getting a certificate will cause problems in the long run, and certainly many students get this, but many do not.
For this reason, it is my opinion that we should look at the current challenges as a “demand side” problem. Now it is not easy for students to outsource their homework to an AI model or agent. Einstein’s innovation—if it works—eliminates the need to cut and paste LLM results. The only way to get readers to learn is to make learning a sound and attractive proposition rather than extracting guaranteed income from a bot.
As I have written before, there are methods that allow for sustainable progress in addressing the challenges of the existence of this technology in ways that are compatible with local and individual conditions. Nothing about the agent AI doing the lessons changes this.
We know what learning looks like—very different from Einstein—but as a program, we haven’t focused on learning as much as we could. Each of these events should be seen as an opportunity to look at what learning really looks like and how what we ask students to do in the classroom supports that learning.
And to reiterate, it’s possible that a recent graduate with a software development background could develop something that works as the product claims, but at this point, it remains to be proven. Many of the productivity AI applications that are emerging on the scene and attracting a lot of attention are more innovative than practical. We need to know if this is true.
Even if this doesn’t work, something like it will probably happen one day, so let’s continue to prepare for that by reducing the importance and relevance of this technology.
