Imagine taking the lightweight, privacy-focused appeal of Linux and completely ruining it by hardwiring AI models straight into the OS. Nobody is switching from Windows for more forced AI features.
The current implementation of it is quite awful. The companies source their data unethically, use fossil fuels to power their data centres, etc. It’s a long list of problems.
The current implementation doesn’t have to be the only one. I think there are better ways to do this. The tool is still neutral to me, but the application isn’t. A knife can be used for slicing tomatoes or murdering people. That doesn’t make the knife an evil tool.
LLMs can be made in better ways too. Currently, they aren’t, which means that in their current state nobody should be paying for LLMs. If these companies start acting more ethically, which they won’t, they could actually earn some of that money.
And then there’s the usage side as well. We’re already beginning to see what happens when you use an LLM in a wrong way. The damage is pretty clear. However, that doesn’t have to be the only way to use an LLM. There should be rules, regulations, guidelines and best practices about this thing. Currently, there are none, so people do just about anything they can think of. Most of it is pretty bad, but that doesn’t have to be the only outcome.