Companies are training LLMs on all the data that they can find, but this data is not the world, but discourse about the world. The rank-and-file developers at these companies, in their naivete, do not see that distinction…So, as these LLMs become increasingly but asymptotically fluent, tantalizingly close to accuracy but ultimately incomplete, developers complain that they are short on data. They have their general purpose computer program, and if they only had the entire world in data form to shove into it, then it would be complete.
That’s easy. The people profiting from it are pushing it hard.
And other companies who had something half-baked just threw it out to both say “me too!” and to ingest as much user input training data in order to catch up.
That’s why “AI” is getting shoved into so many things right now. Not because it’s useful but because they need to gobble up as much training data as they can in order to play catch up.
Going further, they’re like magic. They’re good at what takes up a lot of human time - researching unknown topics, acting as a sounding board, pumping out the fluff expected when communicating professionally.
And they can do a lot more otherwise - they’ve opened so many doors for what software can do and how programmers work, but there’s a real learning curve in figuring out how to tie them into conventional systems. They can smooth over endless tedious tasks
None of those things will make ten trillion dollars. It could add trillions in productivity, but it’s not going to make a trillion dollars for a company next year. It’ll be spread out everywhere across the economy, unless one company can license it to the rest of the world
And that’s what FAANG and venture capitalists are demanding. They want something that’ll create a tech titan, and they want it next quarter
So here we are, with this miracle tech in its infancy. Instead of building on what LLMs are good at and letting them enable humans, they’re being pitched as something that’d make ten trillion dollars - like a replacement for human workers
And it sucks at that. So we have OpenAI closing it off and trying to track GPU usage and kill local AI (among other regulatory barriers to entry), we have Google and Microsoft making the current Internet suck so they’re needed, and we have the industry in a race to build pure llm solutions when independent developers are doing more with orders of magnitude less
Welcome to the worst timeline, AI edition