Stable Diffusion Can Generate Images From Brain Activity

https://sites.google.com/view/stablediffusion-with-brain/ >We demonstrate that our simple framework can reconstruct high-resolution images from brain activity with high semantic fidelity, without the need for training or fine-tuning of complex deep generative models. >We reconstructed visual images from functional Magnetic Resonance Imaging (fMRI) signals using a latent diffusion model named Stable Diffusion. >We also present an objective interpretation of how the text-to-image conversion process implemented by an LDM incorporates the semantic information expressed by the conditional text, while at the same time maintaining the appearance of the original image Your thoughts are no longer private. Thoughts?(USER WAS WARNED FOR THIS POST, You should always elaborate on your opinions rather than just spilling the soykaf.)

>>67566 Good thing all I ever think about is white girls getting BLACKED. It's in a lab setting, and there are other indicators that facilitate thought recognition IRL better than this using analytics metadata. If all it is doing is taking a mental image, you could easily defeat it by picturing something innocuous or irrelevant. Imagine not being in control of your own mind, Lmao. There was a big nothing when thought-to-text was unveiled, but I guess it's good for cripples and people in a vegetative state who can communicate once the technology improve. It might even pave the way for non-body communication.

It's a toy. The entire AI field right now is just building toys. It can generate an image. It can generate text. It can generate a song. It's a confidently incorrect chat bot. Please tech bros can you build something actually practical? Something that we can use to make our lives better? We have enough toys. Where are self-driving cars? Where are new medicines created with the help of artificial intelligence? Where are AI assisted breakthroughs in physics/chemistry/biology?

>>67570 Bettering humanity introduces unforeseen variables and market volatility. Consumer profit is driven by stability. People want escapism, not real change.

>>67570 its putting boxes around all sorts of things, especially for mass surveillance. thats uhh, practical right ?

>>67570 >Where are self-driving cars? Where are new medicines created with the help of artificial intelligence? Where are AI assisted breakthroughs in physics/chemistry/biology? It's either too early or there's not enough interest to make actual stories about them. I am more inclined to think its the former and I hope it is. I don't feel like being narcissistic this time around.

>>67574 fuuuuuck i meant pessimistic. Now to be ridiculed for it.

Interestingly, we live in a time that technology, now produced by the companies of the consumer market, is sold to the consumer market first and only later is it adopted by institutions and government. Perhaps this is one of the aftershocks of globalism that is just starting to become evident. I wouldn't say technological progress has stopped in favor of the wishes of the invisible hand of the market, but the invisible hand of the market is the biggest agent of change in the world and that is a little concerning.

Was just thinking about that. There are ways of thinking that bypass the limitations of linguistic syntax/semantics and the chains of imagery. Think in those obfuscated ways and there will be no way to really parse anything.

>>67570 > Where are new medicines created with the help of artificial intelligence? Where are AI assisted breakthroughs in physics/chemistry/biology? I'm pretty sure AI has already led to breakthroughs in the protein folding problem. Probably most medicines coming out today were developed with the help of machine learning algorithms.

>>67566 >Your thoughts are no longer private sure, all they have to do is put you in an MRI machine and they can know that you're seeing a horse when they show you a picture of a horse. Amazing.

>>67584 >Viagra is sold to keep peoples peenors hard longer, instead of it's clinical use of helping heart problems. many such cases

>>67570 Purely independent technical solutions are impractical right now since any sort of failure leads to massive distrust, even if the human counterparts are massively less reliable (think self driving on tesla cars). On the other hand, expert systems have had massive success and have helped massively in technical breakthroughs. Computer simulations and digital sensors have also been absolutely invaluable to science. Please stop baiting people.

>>67753 >Purely independent technical solutions are impractical right now since any sort of failure leads to massive distrust, even if the human counterparts are massively less reliable (think self driving on tesla cars). Has anyone ever though published reliable statistics about self driving cars? From what I know, currently the autopilot that tesla has is only recommended on roads that are most likely to have least things that go wrong. There's a reason it's recommended that you stay awake and focused after all when driving them. Even exluding that, I am willing to guess there's too little cars to even make solid statistics off. Compared to everything on the road, self driving cars are a minuscule fraction. which make it likely to skew the results in or against their favor. >On the other hand, expert systems have had massive success and have helped massively in technical breakthroughs. Computer simulations and digital sensors have also been absolutely invaluable to science. Please stop baiting people. This though is just bullshit on large part. There's always this talk of "massive success" coming down to justify the most recent trendy thing. An other related example of this is "the software will eat up the world", or "the current problems are unimaginably complex", which are these terms to justify the status quo but are close to meaningless on closer inspection. It's just there to dismiss people's complaints. A good way to stay sane is to ask what kind of problem something is supposed to solve. And most software solves absolutely no real problems. At best it puts band aid on broken glass that should be fixed or replaced. Example of this are countless. One I had to experience as a teen was a system for automatically scheduling of lessons and classes. Which worked fine till something got wrong. Someone got sick, someone couldn't arrive, etc. Since the system assumed that the schedule it made will be held no matter what, the only way to restructure lessons in case of emergencies was to restructure them all, since the software calculated optimal schedule without the teacher who couldn't arrive that day. Which is obviously too much change to handle, so it wasn't uncommon sometimes to have 3 hours you just had to wait through instead of going home earlier. If they wouldn't have a computer, the teachers would just get paper and scribble a new, slightly altered, timetable which most likely would waste less time. The benefit of that system was really questionable to me. But they HAD to "digitalize", because that's the hip thing to do. This might seem like a nitpick, but vast majority of software is like that and simply shouldn't exist. You would think computers would make everything related to controlling stuff easier to manage. Yet the opposite happens. Because for some reason people decide to collect data like crazy. A few decades ago, you could fill in some small form few times and continue with your life. After all, how much data could you collect and process by using office workers? Not much, yet there's no need to have much of it. Things ran just fine if you collect little of it. But now, making a form and processing it is easy. So you are shoved a (unresponsive) form in front of your face on every corner. Of course, whatever you might possibly get from a system like is directly opposed by the fact that someone managing this thing is now being hit with millions of data, and even if he could read all of it, it wouldn't help his goals in any meaningful way since most of it can be seen as meaningless noise. Of course, now we have a solution to this problem. It's called AI. Even though for it to act "intelligently", as in be able to write, it has to be fed multible libraries worth of text and be trained on a supercomputer to immitate one aspect of behaviour that a creature which can power itself for 24 hours on few potatoes can do by reading less than 400 books (I have learned to read and write and I doubt I ever read text the length of 400 books in my life). Thank god we finally solved the problem of handling information by automating the curation and creation of it. I really am glad this so called solution doesn't just create an other problem of moderating halucinated AI data that can be created within seconds on mass, which is an other problem that needs to be solved. Of course though. There are real problems to be solved by software and it can be useful. Even though I think the state of technolgy we have is laughable and majoriy of computer user interfaces don't beat pen and paper, and what has been mostly happening can in my eyes be summed by the word "bullshit"... and for all the so called advancements (which I barely know of anything radical or useful being invented in last decades) still somehow allow for a virus to spread around the globe 3 years ago as if it was the 19th century. BUT there's real room for software. It should be only done though when we understand the problem and software can be (whenever partly or fully) be a solution for it. All of this written in the futuristic jetson year of 2032