https://www.science.org/content/article/ai-re-creates-what-people-see-reading-their-brain-scans
As neuroscientists struggle to demystify how the human brain converts what our eyes see into mental images, artificial intelligence (AI) has been getting better at mimicking that feat. A recent study, scheduled to be presented at an upcoming computer vision conference, demonstrates that AI can read brain scans and re-create largely realistic versions of images a person has seen. As this technology develops, researchers say, it could have numerous applications, from exploring how various animal species perceive the world to perhaps one day recording human dreams and aiding communication in people with paralysis.
Many labs have used AI to read brain scans and re-create images a subject has recently seen, such as human faces and photos of landscapes. The new study marks the first time an AI algorithm called Stable Diffusion, developed by a German group and publicly released in 2022, has been used to do this. Stable Diffusion is similar to other text-to-image "generative" AIs such as DALL-E 2 and Midjourney, which produce new images from text prompts after being trained on billions of images associated with text descriptions.
For the new study, a group in Japan added additional training to the standard Stable Diffusion system, linking additional text descriptions about thousands of photos to brain patterns elicited when those photos were observed by participants in brain scan studies.
[...] Finally, the researchers tested their system on additional brain scans from the same participants when they viewed a separate set of photos, including a toy bear, airplane, clock, and train. By comparing the brain patterns from those images with those produced by the photos in the training data set, the AI system was able to produce convincing imitations of the novel photos. (The team posted a preprint of its work in December 2022.)
"The accuracy of this new method is impressive," says Iris Groen, a neuroscientist at the University of Amsterdam who was not involved with the work.
I'm wondering how this sort of ability will effect copyright, in the long term, when it becomes possible to extract high-enough fidelity copies of media from people's brains, which they have observed before and remember. If someone views an image, listens to a song, or watches a movie, and then downloads a copy from their brain to share, is that copyright infringement? Is the copy in their head infringement? Will the law determine a percentage fidelity limit?
(Score: 5, Insightful) by Barenflimski on Friday March 10, @12:15AM (3 children)
This kind of stuff makes me want to build a treehouse in the Amazon in western Brazil. I'd happily take my chances with a Howler monkey over the idiots that will hide behind this telling the rest of the world how wrong they are.
Give it 20 years and instead of testifying in court, you'll submit your body to an AI brain scan.
Just wait till the foreign intelligence agencies and governments start using this upon their citizens and citizens of the world.
Oh the humanity.
(Score: 2, Insightful) by Runaway1956 on Friday March 10, @01:24AM (2 children)
Need to interrogate a prisoner/suspect? Plug him in to the AI. A couple stories combine drugs and the AI, but the key is that AI making sense of the images in the prisoner's brain. Got a murder victim? If you can get to him/her quickly enough, plug the victim in, to read his last moments alive. "Quickly enough" varies from one Sci-Fi author to another, but the idea is to catch what you can of the dissipating electrical activity.
Abortion is the number one killed of children in the United States.
(Score: 1, Insightful) by Anonymous Coward on Friday March 10, @01:40AM (1 child)
But if I tell you not to think of an object you might still imagine relevant stuff in your mind even if you're innocent... Of course if you can imagine an object that you shouldn't have ever seen before that might be interesting.
However my opinion of the CIA is low enough that I think such stuff will just create opportunities for more evil to be done. Innocent or guilty they're just gonna waterboard you till you can imagine enough details to prove you're guilty.
This might have some potential. Except for the many cases where if they're "alive enough" to read from, they're not that dead yet and the medical people might still be too busy trying to save the victim to let you hook up anything.
By the time you are allowed to handle the victim you might not be able to retrieve the relevant memories.
You might actually be able to stimulate a dead brain to do a recall but I dunno whether that would be admissible in court or should be. In my experience lots of people's memories are pretty bad - they imagine stuff happening that didn't happen. Also they might be imagining someone else is trying to kill them but they never saw the real killer.
(Score: 0) by Anonymous Coward on Friday March 10, @09:05AM
To make things worse many people can have the same false memories too: https://www.mentalfloss.com/article/585887/mandela-effect-examples [mentalfloss.com]
(Score: 1, Interesting) by Anonymous Coward on Friday March 10, @01:26AM
For "sharing" why should there be a difference from now? If you share a copy of a book or song, the same copyright laws apply no matter how you obtained or created that copy.
The difference might be stuff like - say you watch a movie/song somewhere outside your home but due to tech you now have "videographic memory" then you keep enjoying the copy in the comfort of your home. Is that infringement?
In many countries you're not allowed to record what you watch at the cinema... So any artificial memory could be affected by such stuff. You might be forced to remove any device that can do live high quality copies - so you might have to rely on your own potentially poorer quality wetware recall and not be allowed to do a live transfer - which could potentially be higher quality. Even such stuff is probably limited to what your foveas are covering at every moment - for most humans only a small part of the vision is in high res, the other parts aren't, so you have to "refresh the image" by looking around.
In my country the law says that the copies are not infringing if they are for private and domestic use, so there could be more flexibility for such tech. However if I make too many "backup copies" then the law could start considering the copies as not for private and domestic use.
(Score: 2, Troll) by Thexalon on Friday March 10, @04:16AM (1 child)
So, there's this new scientific study on mind-reading, and it turns out a few dozen people in the world can actually tell what other people are thinking. And if they really concentrate carefully, they can even do it over the Internet to total strangers.
Now, you might think this is all hogwash, but I'm actually one of those lucky few who can get into other people's heads like that. And if I really really focus ...
That's it! You're thinking I'm nothing but a liar who can't read minds!
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 3, Touché) by mhajicek on Friday March 10, @07:45AM
You can tell what I'm thinking over the internet, if I assist by encoding my thoughts in ASCII characters and transmitting them to you.
The spacelike surfaces of time foliations can have a cusp at the surface of discontinuity. - P. Hajicek
(Score: 2) by PiMuNu on Friday March 10, @01:03PM
Extraordinary claims require extraordinary evidence. I don't understand the images at the top of the article; in particular, I don't understand the errors. For example, their Expert System has identified a teddy bear with an orange bow tie instead of a teddy bear with a purple bow tie. So what is the source of this "bow tie colour error"? Is the brain encoding "teddy bear" and "bow tie" and "orange", which the Expert System somehow decodes using black magic?
So we read TFA and do a little digging.
> they each viewed a set of 10,000 photos. A portion of these brain scans from the same four participants were not used in training and were used to test the AI system later.
Ah. So the computer is *not* looking at people's brains and deciding that this is a teddy bear with an orange bow tie. It is looking at people's brains and deciding that this is image 7435. Or image 8746. The computer is able to match one of 10,000 patterns, that it was trained on, to whatever the person was looking at. It's a scam.
(Score: 3, Interesting) by bzipitidoo on Friday March 10, @04:48PM
Though this particular report or at least the click bait title, on AI being able to read minds, is likely a wild exaggeration, I'm guessing mind reading is coming. Self-driving cars have been exposed as less capable than they were made to appear, but that too is only a matter of time.
And so, yes, this shows yet another way in which copyright is absurd. Criminalizes the combination of a mind reading device and memory, learning, and education. While education is more than copying, copying is a big part of it. Copyright never was a great idea, but it's only been the technological advances of recent decades that show just how bad it is.