4 ms·

> So if the technical challenges are solved and the AI acting is quite good, that changes your view of it? That "if" seems to be the running thing with everyth

> So if the technical challenges are solved and the AI acting is quite good, that changes your view of it? That "if" seems to be the running thing with everything AI right now. "If" we can fix hallucinations. "If" we can make it actually think and reason. "If" we can make a general AI system. And so on. "IF" we actually hit that point that these systems can create truly creative genuine work instead of being nothing more than predictive engines than maybe we can revisit this conversation. But the fact is we are not there and we are nowhere near there considering we are still just iterating on the fundamental problem of these just being predictive engines. > Actors have worked alongside CGI characters for many decades, I remember watching Who Framed Roger Rabbit in the late 80s. Yes you are right, but there is a pretty massive difference between knowing what the fake thing will do that is already scripted, storyboarded, etc (ultimately controlled by a human the entire time) vs an AI thing that will be an unknown.

elicash 1y ago
1. Hallucinations have been reduced. There are, in fact, reasoning models. The "ifs" are not as theoretical as you imply. But the reason I raise the question on what happens if the acting is of high quality is because it's important to be clear on what your objection actually is to the thing. 2. Your issue on the "how do actors work alongside AI characters" is that actors won't be able to do a good job themselves because it's not storyboarded enough for them? That AI characters aren't rigid enough?
- nerdjon 1y ago
  > Hallucinations have been reduced. Do you have any studies or data to back this up other than "trust us" from the companies pushing the models? And how much "reduced". It is still a fundamental issue that is just being pushed under the rug since it is inconvenient for them to talk about anymore. > There are, in fact, reasoning models. Simulated (fake) reasoning models. They are still fundamentally LLM's talking to itself which goes back to the fundamental issue of hallucination and a complete inability for anything original. Also this "reasoning" has been shown in multiple studies to fall flat on its face with complex issues because it is not true reasoning. There is no intelligence or "thinking" behind it. > The "ifs" are not as theoretical as you imply. My reason for saying it is because it always seems when someone is critical of claims being made the response is 1 of 2 things. 1. Trying to ignore the concerns and just say that it can already do X, its magic, its whatever. Or 2. It is all about the promise. All about what it may supposedly one day be able to do "if" we can just fix this one thing that isn't somehow a huge fundamental issue.

elicash 1y ago

1. Hallucinations have been reduced. There are, in fact, reasoning models. The "ifs" are not as theoretical as you imply. But the reason I raise the question on what happens if the acting is of high quality is because it's important to be clear on what your objection actually is to the thing. 2. Your issue on the "how do actors work alongside AI characters" is that actors won't be able to do a good job themselves because it's not storyboarded enough for them? That AI characters aren't rigid enough?

nerdjon 1y ago
> Hallucinations have been reduced. Do you have any studies or data to back this up other than "trust us" from the companies pushing the models? And how much "reduced". It is still a fundamental issue that is just being pushed under the rug since it is inconvenient for them to talk about anymore. > There are, in fact, reasoning models. Simulated (fake) reasoning models. They are still fundamentally LLM's talking to itself which goes back to the fundamental issue of hallucination and a complete inability for anything original. Also this "reasoning" has been shown in multiple studies to fall flat on its face with complex issues because it is not true reasoning. There is no intelligence or "thinking" behind it. > The "ifs" are not as theoretical as you imply. My reason for saying it is because it always seems when someone is critical of claims being made the response is 1 of 2 things. 1. Trying to ignore the concerns and just say that it can already do X, its magic, its whatever. Or 2. It is all about the promise. All about what it may supposedly one day be able to do "if" we can just fix this one thing that isn't somehow a huge fundamental issue.