AI
As I write this, it is April 13th, 2026. ChatGPT was released in November 2022, with an estimated 5 million users within a week of launch, slamming the world into the "AI era" that we find ourselves in now. That was three and a half years ago, and in all that time, I haven't written a single thing in my digital garden about AI. I wrote plenty to my friends, discussed it at work, fretted about it with my wife, but nothing here.
If a Digital Garden is about learning in public, the meta lesson is that I should just start writing about something as soon as it interests me, as soon as I have opinions about it. But also, it's time to put that principle into practice and start sketching out some of my ideas about all this crazy sci-fi-but-real-life shit that's been going down between then and now.
I'm just going to make a good old bullet list with questions I want to research answers to and other topics, and then fill these in later when I have time:
One thing I'll address off the bat: Is it capitulation to spit in the face of 70 years of Computer Science research and accept that the current techniques of LLMs should be wholly referred to as "AI"? I'm going to go the GNU/Linux route here. Yes, Linux is just a kernel, and GNU is the suite of programs that made that kernel useful. But in the broad vernacular, Linux is an operating system and has been for years. It's vernacular, that's how it works. ChatGPT is understood by millions and millions of people to be "AI", so we can just go with that.
Other questions I'd like to answer:
- Does training an LLM infringe copyright?
- Is training an LLM on the entire Internet ethical?
- Are the models going to get "better", or just plateau? Do we have enough "internets" left to train them on, if parameters is the only thing that makes them improve?
- Is there an AI bubble?
- What trillion dollar problem is AI going to solve?
- Is AI art real art?
- Is it any good anyways?
- If the answer to the two above are yes, then why do people hate it so much?
- Do people compel themselves to answer no to one of those questions in order to enact an anti-"AI art" worldview?
- What are the philosophical problems around songs you like, but you find out they were made by AI?
- What are the general trends in music consumption in 2025 and beyond that are affected by this?
- Why is there so much 'hate' agains AI/LLMs?
- Is Sam Altman a piece of shit?
- Are local models worth it in 2026?
- Why are AI fruit videos so awesome?
Comments
With an account on the Fediverse or Mastodon, you can respond to this post. Simply visit the post on its original server and leave your comment. It and other known non-private replies will be displayed below. Learn how this is implemented here and here.