Over the last two weeks, I've gotten really into
front-end, privacy-first alternatives for popular websites (ie reddit -> teddit) and I noticed all of them offer the opportunity to self-host.
Went down a rabbit hole and just installed
Docker Desktop on my Chromebook. I previously took a course during Covid lockdown but forgot most of it. There are a few interesting related projects that have piqued my interest that I've ignored in the past, but this could be fruitful if I can allocate regular time to it...Things like a
self-hosted audiobook/podcast server, a
locally hosted large language model that I can fine-tune with personal data or just
hosting things for the world as an instance.
So far, I've managed to connect my Chromebook's command line interface to both
Google's AI model as well as
OpenAI's model. Google's just takes a cookie while OpenAI requires an API key.
You can tell by a lot of the links that I've started taking advantage of Github more. I initially thought that AI/LLMs would be the death of coding, but now I think the opposite might be true. I think that solely coding as a career might be harder to make a living, but on the flipside, I think that more and more people will find coding more accessible moving forward because the ability for AI to explain error messages is a game-changer...
I feel like tech-savvy people take for granted just how terrible StackOverflow is, or just searching online for solutions. I'm hardheaded, so I'm used to burning an hour comparing different websites for answers, but even I'm relieved that I can copy/paste error messages and get an immediate answer. It's definitely not always the right answer I've found (the LLMs never tell you that they need more context, they just make an educated, confident-sounding guess even if they're way off) but it makes troubleshooting generally less offputting.