Published: Thu 19 June 2025
By Stefan
In Learning .
When talking about education and large language models the general sentiment appears to be that they are simply the end of education. I find this quite bewildering since I have exactly the opposite perception.
I certainly do not aim to disentangle the emotional-cognitive mess of the topic1 but rather wish to briefly argue for two perspectives.
1. Free cheating for everyone, yay!
In the short-term large language models are nothing new, they just democratize gaming the educational system.
A piece of circumstantial evidence for this perspective is an article published a few months after the launch of the genre defining chatbot. It describes how ChatGPT is essentially the natural successor to an online service called Chegg. Prior to reading the article I had never heard of the company before but apparently in the pre-llm era it was basically the go-to place for college students to outsource their learning activities.
In line with the general discourse on generative AI the articles on the impact in the field of education are primarily centered on individual gains or losses and short-term phenomena like so-called "cheating".2 I'm aware of two recently published rather comprehensive pieces that show this sentiment. One is an apparently quite extensively researched piece and the other a rather sensationalist piece which nonetheless bears some interestingness, although on a different level.
2. Let's steer this in good direction
In the long-term large language models have huge potential to improve individual learning.
I'm aware that one might perceive this perspective as naïve or even utopian since the same has been stated about things like Wikipedia or open educational resources but I remain stubborn with my optimism.3
Generative AI is not the end of education, to the contrary it's a starting point to reflect about what really matters with respect to humanness.
If I recall correctly Geoffrey Hinton mentioned in an interview that his main motivation for entering the field of artificial intelligence research was not to create artificial intelligence as a means to automate cognitive processes but instead to better understand what actually makes the human mind so human.4
It might be obvious, but I would argue that most important prerequisite to successful learning in the age of AI is to develop a healthy learning mindset. That doesn't mean that practical skills are irrelevant, but in my experience these can mostly be acquired in a short amount of time whereas developing a suitable learning mindset can take many years.
A good starting point is to aim for consequently ignoring any systemic incentives and instead taking personal responsibility for one's learning. In practice of course it's often about finding a good compromise to balance out different goals in life, but having an awareness of one's doings and making conscious choices is certainly very different from simply following the path of least resistance.
A mindset of critical curiosity is also certainly helpful. Scientific research and the open source community can be good environments here. There is maybe a bit of luck involved of having the privilege to meet people that have a positive influence in this respect.
A major challenge in forming an opinion on the impact of artificial intelligence on society is that many articles published on the matter do not have an genuine interest in producing insightful perspectives but are rather focused on reproducing the usual tropes. There are very few venues that stand out. In the short-form The Verge is definitely recommendable. For the long-reads MIT Technology Review has some excellent authors. Occasionally The New Yorker also has an interesting piece but there tends be a bit of doomism bias.
I will not go down the rabbit hole of properly defining and evaluating the meaningfulness of this informal term and it's more formal equivalents
I was recently quite delighted when discovering the Open Textbook Initiative of the American Institute of Mathematics
On my book pile currently is The Book of Minds