During my biomedical engineering undergrad, I messed around with some electroencephalograms. They have a long way to go before the aforementioned capabilities. I’m interested, of course, but I have to admit I’m not too hopeful. The current state of the art is like a single unreliable rotary encoder.
Maybe something realistic will be along the lines of reliable single, dual, or triple controls. Imagine a programmable keyboard with 3 switches. I’m just ideating, and don’t really have much empirical basis, but that intuitively seems like a reasonable outcome. If all I had to do was wear a rechargeable (non-bulky) headband to get an extra 3 key-switches, that would be very useful. (Especially for people or applications without use of hands.)
I have no idea how well (if at all) the communication works in the other direction. This is even more fun to fantasize about.
These sort of controllers have already been repeatedly attempted for use in gaming for the past decade or more. While working examples do exist and are sold commercially, they just aren't practical, so they haven't caught on.
So I wouldn't have high hopes for this yet, because clearly the technology just isn't there yet to enable high-fidelity brain noninvasive brain interfaces. But it'll never get there without attempts and funding, so it's good to see DARPA work on this.
Personally I'm more excited for commercial applications of things like the synchron implant (https://synchron.com), but those are pretty damn far from viable as well. And also I'm biased because I want to control everything with my brain at all times.
So I wouldn't have high hopes for this yet, because clearly the technology just isn't there yet to enable high-fidelity brain noninvasive brain interfaces.
I feel like denoising those types of signals would be a good task for AI.
4 comments
[ 2.6 ms ] story [ 18.6 ms ] threadMaybe something realistic will be along the lines of reliable single, dual, or triple controls. Imagine a programmable keyboard with 3 switches. I’m just ideating, and don’t really have much empirical basis, but that intuitively seems like a reasonable outcome. If all I had to do was wear a rechargeable (non-bulky) headband to get an extra 3 key-switches, that would be very useful. (Especially for people or applications without use of hands.)
I have no idea how well (if at all) the communication works in the other direction. This is even more fun to fantasize about.
So I wouldn't have high hopes for this yet, because clearly the technology just isn't there yet to enable high-fidelity brain noninvasive brain interfaces. But it'll never get there without attempts and funding, so it's good to see DARPA work on this.
Personally I'm more excited for commercial applications of things like the synchron implant (https://synchron.com), but those are pretty damn far from viable as well. And also I'm biased because I want to control everything with my brain at all times.
I feel like denoising those types of signals would be a good task for AI.
Semantic reconstruction of continuous language from non-invasive brain recordings (https://www.biorxiv.org/content/10.1101/2022.09.29.509744v1....)
Collaboration with Neurospin, Inria, and Meta (https://about.fb.com/news/2022/04/building-ai-that-processes...)