AI is Evolving Faster Than You Think Pt. 2 (Art and Beyond)

Comments (5) on “AI is Evolving Faster Than You Think Pt. 2 (Art and Beyond)”

1. as has been said before, ai is merely a pattern generator. on that basis, it’s just a generated item, much like a random number generator throws random numbers back at you, sometimes with parameters and noise patterns, such as minecraft’s biome selection that tries to make sure deserts don’t border tundras. similarly, hands are shaped a particular way, and fit in particular spots, and thus, with constraints, is a generatable item. but the key part here is that there are dumb tricks like diffusion in use to help that pattern generation, rather than an actual, honest mimicking of the human brain and its processes, including things like rationale (significance of aspects in artwork and such) that would hold value to stuff like neuroscience. and truly, if we were to not try to blindly generate patterns and simulate, but instead emulate how the human brain is doing things at a decently low level, not only might we save tremendous amounts of power, we could also get art that is made, end to end, exactly as a human would execute on it, as well as the actual application and benefit for neuroscience, and possibly psychology and psychiatry as well, and in a way, it’ll be a symbiotic relationship between the sciences and art, similarly to how the relationship between emulator developers and video game/system preservation is pretty much hand in hand; they both help one another tremendously.

   case in point: byuu/near, with higan. pcb pictures were originally meant for emulator developers, while emulator developers themselves ran into bugs that could only be solved by investigations into hardware. these very investigations, aforementioned pcb pictures included, is exactly what then allowed snes emulators to emulate every single coprocessor ever released for the snes in silicon form to be emulated down to cycle level accuracy. transistor level accuracy will take a ton more horsepower, but cases of this already exist for some far older systems as proofs of concept, despite the pretty much zero gains. that’s not to say it’s useless; if this is achieved at reasonable speeds for things like the motorola 68000, then there will never, ever be a need to worry about instruction timings or related woes in emulators since the transistors will sort that out by themselves in emulation.

   this approach, if translated to artificial intelligence, would be absolutely amazing to see and have, and its implications would be far more profound than a few generated pattern shitposts on the internet thanks to the symbiosis that would exist between artificial intelligence and neuroscience.

   we could’ve had so much more, but alas, greed for time and shortcuts got in the way of all of that.

   Reply

2. Can aps “feel” sympathy, empathy and compassion?

   Reply

3. From my perspective as a person that owns a business that works entirely in the real world and struggling to find professional work hand AI is going to free a lot of working people. It sounds bad and it is bad when people loose their work, but ITs, Digital Artist and overall people that “work” behind the screen are too many in numbers, but at most are not the working force that moves the world. The people that work in the heavy industry, drivers, infrastructure workers and etc. are those moving our economy and they’re on the verge of extinction, before AI is ready to take their job and most people doesn’t even realize it yet, which makes it even worse.

   Reply

4. Shameless pro-AI video. lol

   Reply

5. A year later, this evolution has not slowed down at all. Now not only are the images even more photorealistic, but Sora provides photorealistic videos. Pandora’s box opens more every day. By the time you are reading this, I’m sure even more insane progress has happened. This is what the curve of exponential growth feels like.

   Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.