>What if an AI could interpret your imagination, turning images in your mind's eye into reality? While that sounds like a detail in a cyberpunk novel, researchers have now accomplished exactly this, according to a recently-published paper.
>Researchers found that they could reconstruct high-resolution and highly accurate images from brain activity by using the popular Stable Diffusion image generation model, as outlined in a paper published in December. The authors wrote that unlike previous studies, they didn’t need to train or fine-tune the AI models to create these images. The researchers—from the Graduate School of Frontier Biosciences at Osaka University—said that they first predicted a latent representation, which is a model of the image’s data, from fMRI signals. Then, the model was processed and noise was added to it through the diffusion process. Finally, the researchers decoded text representations from fMRI signals within the higher visual cortex and used them as input to produce a final constructed image.
>In the past, we’ve seen other examples of how brainwaves and brain functions can create images. In 2014, a Shanghai-based artist Jody Xiong used EEG biosensors to connect sixteen people with disabilities to balloons of paint. The people would then use their thoughts to burst specific balloons and create their own paintings. In another EEG example, artist Lia Chavez created an installation that allowed the electrical impulses in the brain to create sounds and light works. Audiences would wear EEG headsets, which would transfer the signals to an A/V system, where the brainwaves would be reflected through color and sound.
>With the advancement of generative AI, more and more researchers have been testing the ways AI models can work with the human brain. In a January 2022 study, researchers at Radboud University in the Netherlands trained a generative AI network, a predecessor of Stable Diffusion, on fMRI data from 1,050 unique faces and convert the brain imaging results into actual images. The study found that the AI was able to perform unparalleled stimulus reconstruction. In the latest study released in December, the researchers revealed that current diffusion models can now achieve high-resolution visual reconstruction.
https://www.vice.com/en/article/wxje8n/researchers-use-ai-to-generate-images-based-on-peoples-brain-activity
How will society function once we reach the ability to read each others thoughts? Everything you can imagine on a page for all to see. Everything. The socialisation of the mind.
It's the last frontier of the surveillance society. I don't think the technology will exist in our lifetimes, but even Orwell predicted it (O'Brien could read Winston's thoughts so acutely that it couldn't just be intuition).
>>28360 >It's the last frontier of the surveillance society
I wonder if it will break from what we traditionally conceive a surveillance society as. Actually going the opposite of 1984 where information is everywhere. It'll probably never happen but imagine being able to peer into the thinking of decision makers as we've come to expect from AI explaining how it came to its responses. I'm sure the racial bias in AI will be overshadowed entirely by what goes on the head of some.
I'd also hazard that everyone around you is much more fucked up than you can possibly imagine. It may well be a curse to know.
>>28364 It has occurred to me that my deepest darkest secret, the sex one, is pretty vanilla really, and most of my thoughts apart from that one are okay to share. I think this puts me quite a way ahead of most people if we were all to have our minds read. There are certain to be other people so much more interesting than me that I would totally get away with anything and nobody would bother reading it.
>>28366 I was thinking that earlier today; aphantasiacs may be the last line of defense against the mind-reading panopticon.
Having said that, what's pictured in the mind's eye isn't really thought or intention. It's entirely possible for most to think and act without visualising anything much.
Which is the best image generation AI to use right now lads? I had a fucking brilliant idea for a prompt earlier.
I'd also like to spark some discussion about the issue of AI and copyright, because it's been a hot topic in the circles I hang about with lately. I am referring to artists who stand to lose out on all that juicy commission income when you no longer have to pay them to draw your femboy fox fursona getting railed by fourteen wolves.
See, the thing is, I don't see the issue from a copyright point of view. It's never been illegal to look at someone's art, and when you put it up on the internet, you are granting license for people to look at it. The argument is that the AI is essentially stealing your art by learning from it, and the debate is about how much of a "new product" the AI is creating when it sharts out an algorithmically generated blend of everything it's seen. It's a real rabbit hole to get into if you want to tackle those questions about what even constitutes originality.
But the things is, I literally don't see how it's any different from any human artist. Point blank, the only difference is this is a computer doing it. No human artist can say they are free from influence. No human artist can say they haven't copied some aspect of an another artist's style, we all have those we aspire to imitate. No musician can say they are not inspired by the amalgamation of the thousands of hours of music they have listened to. We've all stolen a guitar riff or two, then changed a couple of notes later on when we realised.
It seems to me that the issue these people really want to discuss, is one of labour rights. Now that the penny has dropped that they face the prospect of being replaced, of being made obsolete, like the factory worker, the cashier and the file clerk before them, they're suddenly shitting it. But they are too (culturally) middle class and naively liberal to admit that's what it is, so they have to pretend it's something much more petit-bougie like copyright.
>Which is the best image generation AI to use right now lads?
Midjourney gets the best results with the least amount of hassle, but it's only available through a $10/mo subscription. Playground AI and Mage Space aren't as good, but they have free tiers. If you're serious (and you have a decent-ish graphics card), you'll probably want to learn how to use Stable Diffusion with the Automatic1111 GUI. There are loads of tutorials on YouTube and it isn't massively difficult, but you do need to set aside a couple of hours to figure it out.
I totally agree with your position. The way that generative AI models learn to create images is fundamentally no different than how humans learn by example. The trained model doesn't contain a copy of anything. There will be bickering in court, but ultimately the argument that AI is copyright infringement is doomed to fail.
A lot of middle-class people with arts degrees thought that, even though they didn't earn as much as their peers who went into computing or finance or whatever, they were essentially part of a privileged and permanent elite because "computers can never replace human creativity". They might be struggling to pay the bills now, but they'll be laughing when all of the people who do lucrative but boring jobs are replaced by an algorithm. Unfortunately, this world view is contradicted by Moravec's paradox.
Now we've discovered that painting or journalism is actually much easier for a computer to figure out than, say, cleaning toilets or laying bricks; they're doing their nut, because their world view has been shattered and they're facing a profound threat to their status.
As a skilled manual worker, I have very little sympathy. I don't want to get all "learn to code" or anything, but skilled manual workers have faced continual erosion of their working status for centuries. Some of us have adapted and found ways of successfully working in collaboration with the machines, but plenty of others ended up on the scrapheap. Losing your job to a machine was just the natural order of things when it only happened to blokes with calloused hands, but now that people with degrees are facing the same fate we're supposed to believe that it's the greatest crisis facing humanity. These people are going to learn the hard way that their supposedly privileged status was simply a fiction that capitalism doesn't recognise.
> I don't want to get all "learn to code" or anything
And please don't. One reason why there's so much shit code out there nowadays is that plenty of people who are obsolete in their line of work get sent to do some half arsed job centre course scheme for a few weeks where they learn to code just enough so that they'll be able to slap together some third-party libraries while truly understanding fuck all about them.
There's a difference between coding literacy and software engineering. You don't always need the latter, far from it, but it's much easier for somebody who spent years becoming qualified in software engineering to throw down a few lines of efficient code that does what it's supposed to do, than it is for somebody with a passing grasp of programming languages to understand the ramifications of what they're doing.
Cunts.
Just use midjourney if you want the pretty pictures, pay the devs for their great job monetising other people's digital imagery output.
That's the real indignation around it, of course the software is just learning the patterns in a way that's similar to a person, but like a person it needs training data and it's basically just stolen all that for the profit of the fucks that wrote it.
That and it can do what takes a person years of reflection, expertise and skill to learn and just imitates it in seconds.
Whatever. I work in a factory so fuck off with your idiot middle class labour wank fantasy.
