Neuralink—Brain-Computer Interface

On 2022-11-30 Neuralink held a “Show and Tell” event to report on progress toward its goal (from the mission statement) to “Create a generalized I/O platform for the brain”. Elon Musk provided an overview of the company and its goals, then technical leaders explained the details of the technology and hardware being developed. This, like Tesla’s AI Day, is not a press event but primarily a recruiting venue where the company discloses its technology and goals in the hope of motivating those with the talent to achieve them to join the company. The entire event is two and a half hours including a question and answer session at the end.

If you don’t have the time to watch the whole thing, CNET has distilled highlights of the demonstrations to the following ten minute video.

9 Likes

2 Likes

I have renamed this topic “Neuralink—Brain-Computer Interface”, to serve as the place for all things Neuralink and cyborg technology related.

1 Like

https://www.reuters.com/technology/musks-neuralink-raises-280-mln-latest-fundraise-2023-08-07/

Aug 7 (Reuters) - Elon Musk’s brain chip startup Neuralink has raised $280 million in a funding round led by Peter Thiel’s Founders Fund, months after securing approval for its first in-human clinical trial.

The company did not reveal the valuation at which the funds were raised. In June, Reuters reported that the company was valued at about $5 billion after privately executed stock trades.

“We’re extremely excited about this next chapter at Neuralink,” the company said in a post on Musk-owned social media platform X, formerly known as Twitter.

1 Like

I am skeptical about the claimed ability to restore sight to those who are congenitally blind. Though my study was long ago and perhaps dated, one of the basic principles of CNS development was the necessity of specific sensory input to cause the structural development of the required pathways. In all the cases of which I was aware, their absence at birth was not later remediable. Later development was not possible as neuron potentiality decreased over time as development proceeded. Maybe one day genetic manipulations may permit this, but today it is fanciful at best.

3 Likes

image

image

Here is the study brochure [PDF].

image

1 Like

Thanks, but I think I’ll pass on the brain implant:

https://www.extremetech.com/science/new-cap-uses-ai-to-read-minds-and-convert-thoughts-to-text

The closing sentence of that article hit the nail on the head:

Putting on a swimming cap outfitted with EEG electrodes is also far easier to stomach than having an implant surgically attached to your brain, especially when one of the said implants is the brainchild of an erratic billionaire with a dubious sense of responsibility.

4 Likes

We need to compare apples to apples here. So it would be interesting to have a measurement of bandwidth compared to 1) voice 2) keyboard typing 3) handwriting 4) screen tapping and swyping.

1 Like

An EEG based interface is most likely not comparable to an invasive one, but I don’t really know if that is the question. I think the question is how many would be willing to adopt the latter versus the former. From a marketing perspective I think there is no question that EEG would win hands down.

2 Likes

The average typing speed is around 40 words per minute (wpm). What kind of wpm can one expect with an EEG interface? Or with invasive brain sensors?

1 Like

Elon Musk said at the last Neuralink “Show and Tell” event (see the post here “ Neuralink—Brain-Computer Interface” on 2022-12-22) that the invasive Neuralink interface, when fully developed, should allow faster computer input than typing. I would be astonished if you could get any comparable bandwidth or resolution from an outside-the-skull EEG-like technology. While the Neuralink probes allow monitoring groups of neurons in a selected region of the brain, an EEG is looking at bulk effects involving billions of neurons. It’s kind of like trying to figure out what a computer is doing by listening to the interference it generates on an AM radio sitting next to the CPU. You can detect gross patterns in behaviour—for example whether the system is in the idle loop or busy processing, but you have no access to the fine-grained information as to what is going on inside.

3 Likes

The first applications for Neuralink are expected to be neural prostheses to allow nerve control of prosthetic limbs and to restore (initially, very rudimentary) sensory input to those with certain kinds of blindness or deafness. In such cases, some form of high-resolution interface to the affected neurons will almost certainly be required, so those wishing the benefits of the technology will have no alternative to the implanted interface.

I suspect the plan is to use the experience gained from prosthetic interventions (which will face a low threshold for experimental use approval given the potential benefits to those receiving them) to develop the technology for other applications such as eventual high-bandwidth brain-computer interfaces many years in the future.

2 Likes

A significant barrier to widespread adoption of brain implants is risk of infection. Infections in the central nervous system (CNS) are highly morbid. Implants anywhere in the body represent a nidus for possible infection. This has proved manageable for many implants, notably metal joint replacements. One tactic of prevention is prophylactic antibiotics whenever a sub-clinical bacteremia is expected - like oral surgery or colonoscopy. Absent foreign bodies like implants, such bacteremias are usually cleared by the immune system.

Implants in the CNS pose additional problems including cerebrospinal fluid leak at the site where some part of the implant must traverse the meninges - the linings of the CNS, which include dura, arachnoid and pia mater. Such leaks also risk infection by providing additional access to the CNS from bacteremias. Again, most bacteremias do not result in clinical infection, but occasionally do result in unexplained events like abscesses. Even tooth brushing can result in bacteremia. These are hardly rare; these are merely among the known biological risks to the individual.

I go to some length in describing possible complications at the current state of the medical art; That may improve in the future. My rationale is the reigning default absence of critical thinking in most every field thanks to the postmodern (and post fact, post-analysis, post-rational) academy. As we witnessed with the Covid “vaccines” (which do not even meet the former definition of “vaccine”) the desired effects were amplified and trumpeted in Goebbels fashion and even the possibility of unwanted side effects or complications were forcibly unheard. Of course, we are still becoming aware of just what these unintended consequences are, and some are ominous, indeed (like translation of uracil-replaced mRNA into non-functional and/or bizarre proteins; some biologists predict a rise in prion diseases as a result).

I rejoice (maybe a word too strong) at much technological progress like most people, I think. However, to do so absent skepticism, critical thought and continuing honest observation, dissemination, and full discussion of results will foreseeably produce harmful results for both individuals and society at large. I’m responding here to our headlong rush in most every tech advance without requisite prudence. The progressive/globalist/elitist cabal intends to have its way, largely by preventing - by propaganda, censorship and intimidation - informed discussion, especially of topics of the greatest import to the future of humanity. Indeed, brain/computer interfaces of this kind begin to impinge on the very notion of what it means to be human. Combine this with AI and unintended consequences become legion.

6 Likes

It seems that there are some forms of invasive BCI that are less invasive than neuralink (which seems to be in a hurry) or Blackrock, and may stand a better chance of reaching approval for clinical use sooner because of that. Most notably is the one backed by Gates and Bezos

As one that has long struggled with RSI from typing code into a computer for ~30 years, I’m hoping for some kind of breakthrough in non-invasive BCI that might change the equation. While LLMs have shown to be useful, they can’t overcome the hardware limitations. For now, I’ll rely on tools like GitHub Copilot.

3 Likes

AI tools are another great example of an intervention that should be measured in terms of its additive or multiplicative effect on the wpm metric.

1 Like

Something that might hold some promise for non-invasive BCI, not to mention a broad range of medical applications is based on holographic Imaging

The technology is quite fascinating and now it is open source. It makes that little Gadget that Dr. “bones” McCoy carried around in his pocket seem not so far-fetched after all.

3 Likes

image

image

This milestone has not yet been announced on the Neuralink Blog.

4 Likes

Precise control of GUIs using natural language will be available soon: Introducing Adept Experiments

2 Likes