Raw LLM Responses

> it will never be able to infer something beyond the dataset it was trained on. The data itself is one more thing - who has truly unbiased data? It's difficult to create and maintain for sure. And then to top it off since AI is proliferating, it's not long before data from AI is just funneled into training for other AI, and that's gonna be interesting to observe the effects. Not to mention hallucinations. I feel so bad for AI devs because I have this feeling that execs are breathing down their necks to shit out golden eggs, when it may be the fact that LLMs are just not the "path to AGI" or whatever it is these people are envisioning, but rather just a terminal endpoint to the grand tree of AI discoveries. Maybe it's an important step along the way that gets us there and needs more revolutionary inventions in this field. But as it turns out, revolutionary stuff takes ages to come across, since we've picked all the low hanging fruit so it makes sense these things take time. But people want now now now now, fast fast, exponential progresssssss full speed aheadddd!!!!! All this investor money on the balance. I'm sure we'll see things eventually but I also wouldn't be surprised if we plateaued for many years.