If we could tap into someone's brain then we could simply ask a question and find out if they're recalling an event from memory, or whether a statement like "do you know where she is" is truthful. More advanced modelling of someones brain activity can even pick out specific thoughts so there's various ways to extract information from people as this technology progresses.
Although my guess is that none of this is required. An advanced AI model could probably learn your to understand your subconscious muscle / eye movements well enough to extract information without anything more invasive. For example:
1. "Do you know where she is" – AI says he's thinking yes
2. "Is she in the US?" – AI says he's thinking yes
3. "Is she in Texas" – AI says he's thinking yes
And so on until you have the information you need.
> mind-reading computer technology finally works
> "oh no, it actually works - wait, isn't this bad? think of all the ethical implications! what have we done?"
It's not the most dangerous thing that man has ever built, but it is dangerous, and it's probably the most uncomfortably personal.
The following was encouraging:
> In late 2021, the scientists began to run new experiments. First, they were curious if an algorithm trained on one person could be used on another. They found that it could not — the decoder’s efficacy depended on many hours of individualized training. Next, they tested whether the decoder could be thrown off simply by refusing to cooperate with it. Instead of focusing on the story that was playing through their headphones while inside the fMRI machine, participants were asked to complete other mental tasks, such as naming random animals, or telling a different story in their head. “Both of those rendered it completely unusable,” Huth said. “We didn’t decode the story they were listening to, and we couldn’t decode anything about what they were thinking either.”
Maybe one can't shield their true thoughts if they don't know they're in a simulation...?
The doctor came into my room, gave me the option of getting an ICD implant. My first 4000 thoughts were hell no. He explained that otherwise I could just drop dead at any moment, and in all likelihood I wouldn't be the 1/1000 that survived. And if I were to survive I might just be a vegetable next time. I very hesitantly accepted after a lot of thought. BTW this is my 20s that this happened.
Since then more than 1 person has died due to a software error with my device, and I have not been shocked once. I get "patched" all the time to rule out certain errors. But the odds of me dying without the device are still much higher than the odds of my dying with it. More importantly I fully believe the standard of care / investigation into my problems have increased significantly because I opted into having the device. Is that fair? Not really. Is that realistic? Yep, Boston Scientific wants data to make better devices.
I'm in a country where the device was free to install, all care after is entirely free, which definitely helped make the decision. But if there's never a situation in which you'd consider such a compromise I'd say you've probably never woken up in the ICU with no memory of the past 20 days.
There were some follow up studies to the visual decoding thing which is getting more sophisticated w/ diffusion models and other ML techniques improving the accuracy of semantic and visual coding - I think a few were presented at NEURIPS this year
spooky thought - using the output from a Huth-style semantic decoder into the Eddie Chang-lab/UCSF vocal synthesizer?
2) I would say, the neuroethical part of things is growing in importance. What's available in the lab still is way more sophisticated than the devices out in the field - DBS, rTMS, ambulatory EEG, but it's worth having that discussion.
3) Given the above - almost a decade after Obama BRAIN - the lab demonstrations are getting better and better...but none of them have joined DBS, rTMS, ambulatory EEG as something in routine use - i.e. I cannot order an fMRI decoder on a patient in any practical shape or form, or get it covered by Medicare....
I found the article a bit puzzling as it made the researchers seem almost panicked about the neuroethical part. This was either: (1) a journalistic technique (2) the capabilities are progressing far faster here than I have seen, and the article didn't really do a good job of going into it.
I would like it to be based on real progress, but I don't think the article really described the recent progress well.
That oughta hold off the brain parsers for a while.
