Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Multitasking AI assistant concept collage art. AI brain computer capable of doing multiple things at … [+]
For so long, we’ve been hearing science-fiction stories about people receiving brain implants that will impact their psychology, supercharge their abilities, or both. However, we never really had actual instances of this kind of experimentation, until just about a year ago.
That’s when the news came out that NeuroLink implanted the first cortical implant in someone’s brain – and now that those floodgates are open, it makes sense to ask how this type of technology is advancing.
The brain computer interface is, in some ways, a simple concept. But the implementation is extremely complex.
How do the electrodes read the brain activity? How many connections does it take to actually be able to sample someone’s thoughts in a profound way?
In a TED talk this past spring, Danial Hosseintabar talks about some of the math and models behind creating complex results off of diffusion models.
Then he gave an example of how powered AI agents could help automate some of a person’s thought process.
“Let’s say you’re reading through some articles, or you’re scrolling through your favorite social media app, or maybe you’re just checking your emails, and you have received a bunch of emails, and you have to go through each one of those (pieces of) content, you have to consume each of those (pieces of) information for a long time, in order to just get some small part of it which matters to you the most,” he said.
“So what if there was a small copy of you, something that knows the distribution of your thoughts, … and could just give you the things that matter to you the most? For example, the algorithms behind the social media app are usually designed to keep you on the app longer, not to show you things that actually matter to you intellectually. So you can use another AI agent to filter the map, or in terms of emails or messages, you will be able to see those emails with those text messages which matter to you the most, and ignore the ones that you don’t care about.”
Then he went further.
“Imagine that everybody had one of these AI models,” he said. “Each of these models has learned the distribution of their thoughts, which is a problem distribution. And right now, we have a tool in mathematics which can measure how two distributions are similar to each other. So before you even speak to somebody, you can know if they have similar thoughts to you. If they (do), you’re on the same wavelength, (or if they don’t) you’re not. This can actually revolutionize communication.”
He alluded to the BCI as a matter of course, and talks about how it’s been implanted.
“It connects the outmost layer on your brain called the cerebral cortex, which is responsible for capturing high level thoughts, and it can sense the neural activity and send it, over Bluetooth, to a laptop or something else,” he said.
This slide showed some of the theory behind the creation of diffusion patterns and how it applied.
Here’s some more of the math…
So what if you want to know who has become the first guinea pigs in this type of trial?
You can get individual articles on instances like Noland Arbaugh, a 30-year old quadriplegic, receiving an implant, or Brian Bussard getting an optical implant.
But when you ask ChatGPT exactly how many people have NeuroLink implants, it doesn’t come up with a solid number. Nor does it respond with the total number of people who have cortical implants.
However, the model points out that a rough estimate is under a few hundred people at this point.
We can think about the future of mass adoption, and try to understand how these devices will be used, or look at FDA processes for implants like this.
Hosseintabar’s talk and other resources help show us some of the theory behind how this works.
“Even if any of your loved ones pass away one day, you can still be able to (bring) information to their consciousness, or generate new thoughts from them, which is unbelievable,” he said.
That’s one use case, and probably one that is going to be important in many of our societies. But there’s more too, in medicine, in behavioral science, and in the pursuit of longevity and quality of life for people. We’re used to thinking of AI as something that will “take over,” as something scary (like SkyNet) but it could just as easily be a real boon for humankind.