A reflection on the question: What are your thoughts about AI’s impact on education? Written as if I were responding at a fireside chat.
I’d ask ‘what are people already stitching together that a company hasn’t built’ and I think the answer is ChatGPT pairing with YouTube. Youtube is to lectures as chatgpt is to office hours. And if we don’t think about education as a system and think about it as a simplified term for increase knowledge of the human, this does the job. You can test yourself with ChatGPT. I frequently ask it to give me programming problems and I work through them, submit my answer and get feedback. I could arguably do this without ChatGPT via a textbook and its practice problems but that’s in a static state where the “tutoring” is already in the textbook and it’s not changing. ChatGPT is in an active state because the “tutoring” evolves as I ask more questions and go deeper on a concept I missed. More importantly, ChatGPT allows me to ask why and get an answer instead of asking why and writing in the course forum or googling. Online courses haven’t worked for several reasons but I think the inability to contact a teacher through an OH session is one of the non-obvious reasons.
The question could arise, how do we credential this learning? It’s happening off-campus so isn’t it impossible to award credit hours that contribute to a degree. Youtube is changing this (https://bit.ly/3XVpdfS) and the concept of crypto credentials or authority verification or peer-to-peer verification also exist. They’re non-traditional forms of credentialing for education but traditional for developers. For example, authority verification is getting a maintainer on an open source project to approve 5 of your pull requests. Your contribution, if meaningful, can be shown on your GitHub repository with public view of the PR sent and approved w/comments of “good code” or “clean functions.” Fundamentally, a credential verifies knowledge. To meaningfully contribute to an open source project, we can assume a developer has knowledge. This isn’t equivalent to a CS degree but should be valued more than it currently is. Peer-to-peer verification happens in hiring developer talent into companies. For example, my friend Nicolas Gatien (he was in grade 11) was hired to work as a python developer for a startup and the only thing the employer reviewed was his GitHub. No grades or school credential, just how clean his code was. These activities happen off-campus without the adoption of traditional education but they are a form of credential and some in SV might agree it matters more than an on-campus credential.
I don’t agree that credentials don’t matter or that companies shouldn’t look at a candidates degree when they’re hiring. The context of the degree should matter but if they went to a school, majored in art but spent all their time coding front-end projects that would be an interesting fact to know; a fact that would be absent without looking at the context of that candidate. I think the school is somewhat important for the context but less important to price that person’s market value.
Market value is the core thing people should think about when applying to college. If they believe a college will increase their market value, then attend. If not, move to a place where you can do what you’re best at and build up your value there. In some career paths, experience is more valuable than a college degree. For example, being a Chief of Staff requires more interaction with people in a specific field or company and is more valuable than a Bachelor’s in communications.
There is evidence that certain degrees have negative ROI and for those degrees college doesn’t make sense. And it’s likely that people remain interested in career paths with a negative ROI for college. I have a friend who is interested in creative writing and pursuing a major in it but spends a few hours outside of her major taking math and accounting classes. She’s interested in the material but more interested in having a backup skillset that’s more monetizable than the one she’s in college for. I don’t believe this is a problem of the university. It could partly be the typical regime change that comes with technological revolutions. But I think it’s the economics of writing that make it so difficult to monetize that career as many others. I've written before about how I think this problem gets solved (https://www.zaynp.com/aim).
Outside of teaching and credentialing, AI changes where human capital should flow. If ChatGPT or GitHub Copilot can program faster and cleaner than a current programmer, the cost to produce and rewrite good software dramatically reduces. So does the number of software engineers required. I think the ChatGPT demonstrates that tech in the real world matters more, ie - hardware. Do we have enough people who can make things with their hands, who can do the precision work robots can’t do yet? Or, non-technical, do we have enough doctors and healthcare workers.
I’ve thought about making a skill market cap that has a list of top skills based on:
It would update on a weekly or monthly basis and everyone would know what is the highest leverage skill. It might end up surprising people too. Even today, doctors and nurses might be higher priced skills than software engineers, although maybe that’s non-obvious to some.
The last point I’ll mention on this question is, the teachers unions have an incentive to keep AI out of the system as long as possible. In fact, it’s a keep all disruptive technology out of the classroom rule. There’s a great book called Liberating Learning with an analysis of the politics behind education and I’ll read you a few quotes to close. The first: "They [teachers unions] have an interest in pressing for reduced class sizes and other means of increasing the demand for teachers. They have an interest in fighting for bigger budgets and higher taxes. And so on."
The second: “This school [Dayton Academies] is one of the pioneers in bringing distance-learning technologies to students who "attend" from locations all over the state of Wisconsin, and whose needs were not being met by their own district. Exciting, right? But also threatening to the districts that are losing students and resources to this innovative school. And threatening to the state teachers union, which, among other things, wants to protect the jobs of teachers in those districts, and does not like WVA's ability to operate at lower teacher-student ratios.”
Neither of these directly mentioned AI but there are references to maintaining the jobs of teachers and fighting against technology because it means less stipends per teacher and less power for the union.
I’d pay close attention to off-campus tech platforms, like Youtube. I think we’ll see my ‘Tech companies will build the future of education’ essay continue to manifest with Google continuing to lead the way.
Ps - An update on my Tech companies will build the future of education essay: