Learning in the Age of AI

A quote from David Weber’s Heirs of Empire:

“With neural interfacing, there was no inherent limit to the data any individual could be given, but raw data wasn’t the same as knowledge, and that required a whole new set of educational parameters. For the first time in human history, the only thing that mattered was what the best educators had always insisted was the true goal of education: the exploration of knowledge. It was no longer necessary for students to spend endless hours acquiring data, but only a matter of making them aware of what they already ‘knew’ and teaching them to use it — teaching them to think, really — and that was a good teacher’s delight.”

This line struck me when I first read it about 30 years ago.¹ The internet was still fairly young but even then I could see that we were getting to a place where all the world’s knowledge was accessible. Now, not only can I look up anything I am curious about, I can ask my favorite bot to summarize it or teach it to me. I may not have the perfect recall that Weber envisioned but with the neural implants that are being developed it is really only a matter of time.

I’ve been using Claude to do so many things. At first I was reluctant because ‘muh agency’ but now I am realizing it is allowing me to make things exist that would not have if I had to rely on my own skills. I am clever, I am smart (whatever that means), and I have good ideas. I know this because I have folders, and folders, of semi-developed ideas that other people have brought to fruition,² but I lacked the time to develop them fully and/or the skill set to see them through. Now I can ask Claude to walk me through it and I am creating things, wonderful things, that would not have existed without it.

No More Drudgery

Raise your hand if you like filling out configuration files. No one?³ Really?

Claude doesn’t mind. Claude remembers how to set up Gradle. Claude remembers the order of things in a Dockerfile. Claude can figure out that this version of Ubuntu is wrong and it can fix it faster than I can figure out what the issue was.

So what good am I? Claude doesn’t come up with new ideas. Claude recombines things in new ways. But that is what creation and artistry is. The limit is new ideas. The limit is WHAT DO WE WANT TO EXIST THAT DOESN’T. I heard a talk once that was lamenting the level of education needed before you got to the cool stuff in math. Well CS has the same limitation. Yes I love white (or green) text on a black background but then I am old and that’s what was “cool” when I was a young warthog, but now we can get JavaFX to display things in a window with buttons and images before we even know what a scene graph is. Maybe we’ve lost something but then I also don’t know how to thresh wheat but I do enjoy me some bread.

No Pollyanna! No!

I don’t think we are entering into a utopia, I fear the human condition precludes that, but I do think the era of being a tech-bro and making stacks-on-stacks because you know how to sort an R-B Tree in O(n log n) time is coming to a close. When my mom can ask a chatbot to make a piece of software, why do we need devs? Because history and context matter. This means we will still have and need tech-bros but, BUT, we will need people who are passionate about learning, exploring, and creating. If you are in it for a quick buck… well… that may not be valid any more.

There are equity issues.

I am fortunate that I can afford the $20 a month for Claude. Not everyone can. I also think that the era of NEEDING Claude is going to be a bit short-lived because my five year old MBP can run LLMs locally pretty well. I don’t think the capabilities of the LLMs are going to continue to improve as rapidly as they have been but I DO think they will become more widespread. So software will change. We will change.

What this means for education

Our roles as educators will become different. I think we will become what we should have always been: people who inspire and guide curiosity. You don’t need to go to school to learn how to paint, play an instrument, or write a book but WOW does it help. Seeing what others have done, learning the context of what works and what doesn’t, learning the WHY and the HOW are instrumental in learning the WHAT.

Back to the quote above: future learners will need to be shown WHAT they know, WHAT they have access to, and what they can do with it. When I learned about basic Newtonian physics in college I immediately used it to make a game with accurate ballistics. The game already exists in several forms (Scorched Earth is the oldest version I know of) but it was still fun and I made it. What did this teach me? That playing a game I created, crappy as it was, was more fun than playing one made by someone else. I enjoyed the process and learned more along the way. This was recombining things but it made me want to learn more.

The future is scary.

The great Yogi Berra said “Predictions are hard, especially about the future”⁴ and wow is that true. I don’t know anyone (not even me) who had world-altering LLMs on their 2023 bingo card. Maybe aliens will contact us and tell us that AIs are forbidden in the Galactic Syndicate or maybe they will show up and tell us now that we have AIs we can join the Federation, so long as the AIs vote for us.

I hope I still have this job in 5 years. I hope I am retired in 23.⁵ I hope that my son grows up in a kinder, more accepting world. Despite all the turmoil and terrible things, Dr. Martin Luther King Jr.’s words reverberate: “The arc of the moral universe is long, but it bends toward justice.” World hunger is down. Poverty is down. Life expectancy is up. We need to work on those bowling scores and being excellent to each other and maybe then we can have peace in our time.

For now go make something that didn’t exist this morning. Make music, art, a game. It doesn’t have to be good it just has to be yours.