AI is starting to make sense

When ChatGPT came onto the scene a few years (!) ago I was initially surprised, then became very quickly annoyed and doubtful - especially when people started to claim that AGI was now just on the horizon. I mean FFS - LLMs are just fancy auto complete, right?

Turns out if you spent some time with what is now available, especially beyond just Eliza-style conversations with chatbots - things start to make sense now.

For the last two weeks or so I have been using LLMs to work on some coding projects of mine. Stuff that I was thinking about for years but never got to. Started with Aider but quickly switched over to the more capable Cline. Eventually I started hitting a wall though where Cline would repeatedly generate 300+ tool calls and then get stuck. Updating to the latest version didn't really help so on a whim I tried OpenCode - and stayed with it. It really has helped me become productive again. I no longer have to code every single line myself, instead I can focus on higher level architecture - stuff where my experience actually comes in handy.

I'll post more about these projects in the near future but those are topics for literally another day.

Oh, and I am still 100% certain that OpenAI, Meta, xAI, Facebook, and all the other big AI companies are royally FCKD. There is no way to recoup all that money spent, ever. Costs are crashing down faster than a lead balloon (although those things do float... hmm... miss Mythbusters...) with AI token generation becoming a commodity.