1) Huth lab and Gallant lab are extremely highly regarding in fMRI encoding/decoding, plus their code and models are released as open source; much admiration for that.<p>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<p>spooky thought - using the output from a Huth-style semantic decoder into the Eddie Chang-lab/UCSF vocal synthesizer?<p>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.<p>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....
> spend decades deliberately developing mind-reading computer technology<p>> mind-reading computer technology finally works<p>> "oh no, it actually works - wait, isn't this bad? think of all the ethical implications! what have we done?"
The ethics of this seem awful. If the tech actually works, it's almost certain that governments will start using it to force information out of people.
There are at least 2 PhD EE teams interested in developing and implanting 1k's-10k's of MEMS IR sensors with an option to add electrical or light stimulation for a complete cybernetic brain interface.
It's amazing how much information can be extracted from functional MRI data. That's like figuring out what a chip is doing by looking at which parts are using warmer because they're using more power.
No way in hell I would ever install a computer in my body, not even a pacemaker. I would rather shrivel up and return to nature than get rekt by a hardware/software mistake.