This post did not contain any content.
-
@Paradox @linear @wafflesies It's depressing. They add more nodes to the neural network & continue training far beyond when diminishing returns manifest. It's beyond massively inefficient. There's also "AI Powered PhotoShop" from Adobe now, & I assume that kind of thing is a big part of this. I can imagine a malicious actor making a normal collage, erasing the seams, then saying "ok shitgpt fill in the gaps with mud for me", or "ok shitgpt make this but less obviously fake".
@jackemled @Paradox @linear this post is an indicator that you already don't know whats real anymore
Bear that in mind moving forward -
@jackemled @Paradox @linear this post is an indicator that you already don't know whats real anymore
Bear that in mind moving forward@wafflesies @Paradox @linear It's depressing, but naturally occurring neural network architectures are capable of self improvement & can also train much more efficiently than artificial architectures. I will always be more adaptable than a LLM or image generator. I now have a sample size of five for this "generation" of random image generators & I'm already ahead of them. I probably will not fall for this more than two or three more times, assuming each time after this only has one image.
This has been a cat & mouse game for a few years now, but unlike what techbros think, I am the cat & they are the mouse. They are the ones who must keep finding new ways to trick me, it's not me that has to keep finding out how to evade their tricks.
-
@wafflesies@infosec.exchange this looks ai-generated or something. each of the images are different, even if they look vaguely similar
also if you think a gas engine would only get 2 hours, even on one of those things, you've probably never seen an actual gas powered laptop
-
@wafflesies @Paradox @linear It's depressing, but naturally occurring neural network architectures are capable of self improvement & can also train much more efficiently than artificial architectures. I will always be more adaptable than a LLM or image generator. I now have a sample size of five for this "generation" of random image generators & I'm already ahead of them. I probably will not fall for this more than two or three more times, assuming each time after this only has one image.
This has been a cat & mouse game for a few years now, but unlike what techbros think, I am the cat & they are the mouse. They are the ones who must keep finding new ways to trick me, it's not me that has to keep finding out how to evade their tricks.
@jackemled @Paradox @linear while I'm not sure about AGI technology will continue to improve and reality will become harder to distinguish
Remember technology can never move backwards only forwards -
@Paradox @linear @wafflesies I explain the artificial image aging in the linked post. Basically they apply lossy compression & lossy data transfers until the image looks like it has been circulating on the Internet for a long time, with people downloading & reposting or screenshotting & reposting several times.
@jackemled @Paradox @linear yeah common ageing thing ive seen it used in deep frying
-
@jackemled @Paradox @linear while I'm not sure about AGI technology will continue to improve and reality will become harder to distinguish
Remember technology can never move backwards only forwards@wafflesies @Paradox @linear It can absolutely move backwards. I don't think Greek fire has been figured out again yet, & no one has synthesized sylphium either. Also random image generators are easily defeated by turning the screen off, so I don't think I have to worry about distinguishing reality from noise. It's not like I'm stuck in The Matrix.
-
@wafflesies @Paradox @linear It can absolutely move backwards. I don't think Greek fire has been figured out again yet, & no one has synthesized sylphium either. Also random image generators are easily defeated by turning the screen off, so I don't think I have to worry about distinguishing reality from noise. It's not like I'm stuck in The Matrix.
@jackemled @Paradox @linear I think in the end were gonna have to innawoods
-
@jackemled @Paradox @linear I think in the end were gonna have to innawoods
@wafflesies @Paradox @linear I haven't heard of it.
-
This post did not contain any content.
@wafflesies you don't have to worry about loud fans
-
@themannofcool @swirly @wafflesies No, it works — see, there is an external battery indicator — gasoline level in the white tank
-
This post did not contain any content.
Dieselpunk
-
This post did not contain any content.
@wafflesies It uses an extremely energy-dense fuel and it *still* only lasts an hour and a half?!? Holy shit.
-
This post did not contain any content.
@wafflesies
Now I want to build this -
@linear @wafflesies
The details are impeccable, but yeah it's weird that the main parts of the engine seem to be organized differently in each one.@Paradox @linear @wafflesies *cool looking image*
conversation : "it's all AI btw"
It's a new daily facepalm I'd need a cultural name for it, something like Aibummer or AI trap
-
@wafflesies
Now I want to build this@ranja here's something a bit more realistic lol
-
@ranja here's something a bit more realistic lol
@wafflesies
It's just not the same without the smell of two stroke oil in the coffee shop -
This post did not contain any content.
@wafflesies Ugh, reminds me of this ad: https://youtu.be/Il8ATz16pyM 🥴
-
P pelle@veganism.social shared this topic
