May 30, 2024

Thursday Open Comments

Let’s return to an old fan favorite: AI and LLMs—but this time with a less enthusiastic take! (Yay!)

The point of discussion is a podcast, and the host is Quincy Larson, the founder of freeCodeCamp. FreeCodeCamp offers what its name says, and it has become an impressive online source for self-teaching coding skills.

Larson is screen right in the video (though not in the title image, confusingly). The guest goes by “ThePrimeagen” and hosts live coding demonstrations. He is an accomplished programmer, to say the least.

A lot of the discussion was interesting, although I found the guest unusually annoying in a several ways related to being very geek-like, almost to a cartoonish degree. I mention that I found him a bit annoying because I might have stopped listening because of it, yet was too intrigued by the conversation. There were quite a few interesting points about using AI assistants for coding and for learning to code. In the guest’s experience, the AI made him realize that he quickly became too reliant on it, and that its help prevented him from learning to the actual programming language.

It’s familiar sounding, in the sense that sat-nav technology has made many of us less capable of using printed maps to find places, for instance. I’d imagine that millennials and younger folks aren’t familiar with using brick-and-mortar libraries or paper-based reference books (dictionaries, phone books), either, and they will miss out on the experience of finding related information accidentally when looking through shelves of books in a library and coming across unexpected titles nearby.

The podcast segment at around 13 minutes contains some interesting tidbits about motivation in learning. Around the 1 hour and 40 minute mark is a segment where they discuss the implications for the job market and the downsides of becoming reliant on AI tools. The speculation seems just as reasonable that AI will lead to more jobs rather than fewer, as much of the fear-mongering has it.

It’s a different critical view from the end-of-the-world proclamations from late last year, when AI burst onto the scene. The risks described sound more plausible and less panicky.

The podcast is a bit of a different discussion from the usual here, and I’m sure most won’t want to listen to the whole thing—it is long. I thought there were some thought-provoking points in it, so I tried to find the locations nearest the interesting segments, as noted above.

Finally, here’s the host’s own description of the episode:

On this week's podcast, I interview ThePrimeagen, a former Netflix engineer who live-streams his coding on Twitch. He shares his thoughts on AI tools and why he ripped GitHub Copilot out from his code editor. He thinks AI will create more software engineer jobs than it destroys. We also explore his love of hard Nintendo games and why he left Silicon Valley to live on a horse ranch in South Dakota.

.