What is Neuralink?: Sewing our Brains into The Singularity

How can humanity ever hope to keep up with the Cambrian Explosion of the AI space?

According to Elon Musk, one way in which this might be done involves a technology called a brain-computer interface. The Towards Data Science blog defines this as a device that lets humans connect with an AI “by brain activity only.” This would likely mean that a team would have to implant some sort of computer chip into a person’s brain for this to even be possible.

Musk has been testing this idea through his new company called Neuralink, which has been almost exclusively working on BCIs since it was incorporated in 2016. As the TDS blog points out, BCIs are not necessarily a new idea, but implanting them inside someone’s brain and having them be AIs is, at least in practice.

For now though, this isn’t something that’s about to happen any time soon. Neuralink has been spending the bulk of its’ time researching the applications of BCIs and similar technologies to the treatment of brain diseases. According to the Wait but Why blog, in 2017, the Neuralink team had the goal of releasing a workable version of something that could be applied to such a use case in four years.

As of now, little news seems to have surfaced on the subject beyond Neuralink testing a prototype of its’ early-stage BCI on rats. In a nutshell, Neuralink has finally publicized some of its’ findings from these efforts. though they’ve done so somewhat anonymously. The easiest way to conceptualize what they’ve created so far is to think about “a sewing machine” that sews a special kind of AI chip that’s constructed of “electrode threads” into place on a rat’s brain.

Judging by the research itself, they’ve experienced everything from long-term success, to the rats undergoing significant damage to their brain tissue as a result of the treatment. Despite these not-so insignificant stumbling blocks, it is easy to see the promising nature of Neuralink’s work, especially when it’s made known that the same US government agency that helped fund the defining research for the Internet is one of their partner’s.

As Bloomberg’s piece on the subject suggests, this is also likely only the initial iteration of the technology. Therefore, it’s reasonable to expect that it won’t be tested on humans until several more rounds of meaningful changes are made.

Publications like TechStartups and the United Kingdom’s Express have panned Neuralink’s efforts as highly dangerous. Steven Woo, who is the founder of a data science firm called Rambus, pointed out that if hackers could access the servers that the BCIs depended on, then they would be accessing the brains of real people.

If BCIs did turn out to have significant security vulnerabilities, then the implications that arose from such a “brain hack,” would be disastrous to say the least. Discussing this at this point is, however, jumping the gun in some respects. For example, who’s to say that BCIs will depend on any sort of centralized servers?

Why not use the Blockchain for the entire, inherent AI system? That would eliminate the risk of keeping sensitive data in one dedicated location. Essentially, the more people that acted as nodes on the network, the more secure it would be, provided that the governance framework was designed in a careful fashion.

Regardless, since the paper that Bloomberg initially discovered this week was the first public record of Neuralink’s findings, anything is possible at this point. Even the blockchain is on the table, until proven otherwise. As this saga develops, expect us to keep you informed but until then, check out our list of suggested resources on the subject below. Furthermore, for more on the technologies involved in BCI-related research, check out “convolutional neural networks.”