Search Results
7/23/2025, 7:41:38 PM
>>716209632
Industry artist here.
Let me tell you about AI.
The problem with AI is that it lacks consistency, and it's hard to make edits with. When it works it works REALLY nice, can save hours of your time. However the issues start to arise when you don't generate exactly what you want, or if you generate something that's close but not quite there - AI struggles to fix those gaps in ways that look natural. You either have to try and edit it yourself (still requires a full creative skillset), or just try again with different parameters and pray the AI eventually hits the mark. So for people with strong visions for what they want, AI ends up missing the mark - but for people who do not care at all, it's perfect.
That's the first major problem with AI. The second problem is that not everyone is an artist, nor really benefits from the image/video generation aspect - and it's largely being pushed towards people who don't really have much use for it. Its predictive text use sky rocketed in quality until it hit a point of being genuinely believable, but then plateued at a point where it still isn't quite good. It still suffers from significant hallucinations, and struggles to retain facts over longer conversations - more often choosing to make up things that sounds correct instead of being factual. On top of that, it's susceptible to some good ol' social engineering, someone with the right knowhow can and will crack open an AI and get it to do what they want.
All of this means that large commercial application for AI is still not quite good enough. It sits in the same bubble as VR. It's cool, but it's just not there yet and it won't ever really BE where people fantasize it as until there's more major leaps in technology.
It's basically snakeoil as it currently is. It can't be much else until it gets significantly better. It'll get better, but the push to make it financially successful RIGHT NOW is too soon.
Industry artist here.
Let me tell you about AI.
The problem with AI is that it lacks consistency, and it's hard to make edits with. When it works it works REALLY nice, can save hours of your time. However the issues start to arise when you don't generate exactly what you want, or if you generate something that's close but not quite there - AI struggles to fix those gaps in ways that look natural. You either have to try and edit it yourself (still requires a full creative skillset), or just try again with different parameters and pray the AI eventually hits the mark. So for people with strong visions for what they want, AI ends up missing the mark - but for people who do not care at all, it's perfect.
That's the first major problem with AI. The second problem is that not everyone is an artist, nor really benefits from the image/video generation aspect - and it's largely being pushed towards people who don't really have much use for it. Its predictive text use sky rocketed in quality until it hit a point of being genuinely believable, but then plateued at a point where it still isn't quite good. It still suffers from significant hallucinations, and struggles to retain facts over longer conversations - more often choosing to make up things that sounds correct instead of being factual. On top of that, it's susceptible to some good ol' social engineering, someone with the right knowhow can and will crack open an AI and get it to do what they want.
All of this means that large commercial application for AI is still not quite good enough. It sits in the same bubble as VR. It's cool, but it's just not there yet and it won't ever really BE where people fantasize it as until there's more major leaps in technology.
It's basically snakeoil as it currently is. It can't be much else until it gets significantly better. It'll get better, but the push to make it financially successful RIGHT NOW is too soon.
Page 1