I'm sorry, are we saying we can use computers to read minds now? Because that's what this sounds like. Yeah, fine, click the link, don't click the link, I'm going to sum up either way: according to a paper published by researchers from the National University of Singapore and The Chinese University of Hong Kong, it's possible, using AI learning technology, to reconstruct images from the mind's eye.
|
Above: the grim, confusing future we're destined to age into. |
"...we propose Mind-Video that learns spatiotemporal information from continuous fMRI data of the cerebral cortex progressively through masked brain modeling, multimodal contrastive learning with spatiotemporal attention, and co-training with an augmented Stable Diffusion model that incorporates network temporal inflation."
-the abstract, the part that's supposed
to be a quick summary of the research
But again,
if I'm following, I think the idea is that they put your head in some kind of MRI and what you see in your head ends up as a thirty-frame per second video file with 85% accuracy. What 85% accuracy means or how that's even measured is a mystery to me, but fortunately this scientific paper
has pictures:
|
Original video on top, video as filtered through a human brain, an MRI machine and then the AI software on bottom. You know, this might now be the most ominous cat video the internet has to offer. |
|
I mean, dang Wen 2018, why'd you even show up to this party? |
Ok, so you might be thinking to yourself "m'eh, it's not that accurate. But bear in mind that this is a computer that reads your mind. And while the kitty below has the distorted, somehow off quality that a lot of AI generated imagery seems to have right now, it is, and I can't stress this enough, coming from a
computer that reads your mind. It's also, as the paper points out, far ahead of previous attempt to--huh? What's that? Oh yes, evidently this has been done before with less impressive results. But still:
mind reading computer. Check out the comparison to the right and ask yourself if you could do any better leaning the spatiotemporal multimodal--sorry, are my ears bleeding?
But what's this even for? you might reasonably ask. I don't know, probably some medical applications? The researchers don't say. It's not really their job I guess, but they do bring up the idea that while brain to computer interfaces are promising, there probably should be some governmental regulation to prevent malicious uses of the technology. But it's probably unnecessary. I'm sure everyone will use the power to record information directly from the human brain wisely. After all, we've had such a great track record up until now.
|
What could possibly go wrong? Everything. Like, just everything. |