The problem with using the for LLMs is that the product itself is questionable in nature
No it’s not. It’s imperfect and flawed, but still very useful. All those companies buying enterprise and team plans don’t do it because they expect to get AGI one day, they do it because it’s useful to them now. I see it in my own company, being used for all sorts of stuff, including software development assistance. Usefulness is what counts, not if it sometimes gets something silly wrong which then everyone shouts out from the rooftops.
Yes, there is possibly too much money in it for what it potentially can deliver within 5, 10, 15 years from now in terms of revenue, but even if the technology stopped evolving today and we would just keep getting faster and cheaper outputs, it would still be incredibly useful and valuable. And of course, it’s not stopping.
I said questionable in nature in comparison to the other two bubbles, which were established technology to build off of.
I agree on usefulness, to a point. It’s being used everywhere, and while an LLM breaking in some situation can be countered, look how many places it’s being given full reins and crossed fingers. Knowing it makes mistakes. Bad mistakes, Maybe not often, but sometimes you need as close to 100% as possible, with redundancies in place.
I use LLMs myself, and in just using it for things that aren’t critical, I’ll catch that small percentage and realize, it’s not good enough for most things it’s being put in. I do find it amazing that it produces what it does, but that doesn’t make me ignore when it breaks gloriously.
Yes, there is possibly too much money in it for what it potentially can deliver within 5, 10, 15 years from now in terms of revenue, but even if the technology stopped evolving today and we would just keep getting faster and cheaper outputs, it would still be incredibly useful and valuable. And of course, it’s not stopping.
I said questionable in nature in comparison to the other two bubbles, which were established technology to build off of.
I agree on usefulness, to a point. It’s being used everywhere, and while an LLM breaking in some situation can be countered, look how many places it’s being given full reins and crossed fingers. Knowing it makes mistakes. Bad mistakes, Maybe not often, but sometimes you need as close to 100% as possible, with redundancies in place.
I use LLMs myself, and in just using it for things that aren’t critical, I’ll catch that small percentage and realize, it’s not good enough for most things it’s being put in. I do find it amazing that it produces what it does, but that doesn’t make me ignore when it breaks gloriously.