The Inevitability of Brain Computer Interfaces

It's only a matter of time


All of our interaction with the physical world is constrained to a single, fragile, bounded interface called the human body. In recent years, however, the notion of separating the function of the brain from the limitations of the human body has gone from implausible science fiction to probable reality—and, I would argue, inevitability as well.

We’re talking about Brain Computer Interfaces (“BCIs”, sometimes referred to as “Smartbrains”). These range from partially invasive to fully implanted machines embedded within the human brain, which allow the recipient to control something (a computer, a keyboard, a drone, etc.) with nothing other than their thoughts.

Far from being the only company pursuing this dream, the most well-known startup in the space is nonetheless Elon Musk’s Neuralink. Just over a month ago, Musk himself claimed (but did not prove) that the company had implanted a device in its first human patient, and that the recipient was able to mind-control a cursor on a screen.

Before being tested on humans, Neuralink was tested on monkeys like this one:

AI’s Impact

Why the sudden inflection point and the inevitable proliferation of this technology? AI, of course.

In the past few years, AI has made advancements in the science and theory of neural implants never before achieved, accelerating the pace of development to new highs (here’s one such example, and here’s another).

It isn’t hard to appreciate why this pairing of AI and BCIs is so powerful.

If Large Language Models (“LLMs”) such as ChatGPT have shown humanity anything, it’s not merely that we can build machines to write, speak, or even “think” like a human. Rather, the implications of LLMs extend far beyond their current applications. What we have learned is that sufficiently large neural networks trained on enough data can learn to emulate physical phenomena that human beings do not currently and may never understand. In other words, AI permits the development of mimicking technology that bypasses the necessity of truly understanding that which is being mimicked.

No one at OpenAI, Google, or anywhere else understands the nuances and complexity of all human language. Yet ChatGPT, Gemini, and other systems they’ve built nevertheless seem to do so.

At its core, a neural network is a way of training a machine to approximate a mathematical function. The more training data and parameters it is given, the closer its approximation gets to reality. Assuming all worldly phenomena are underpinned by some mathematical logic, this means that a neural network at scale could theoretically learn to predict or replicate anything. Ultimately, its trillion weights (like tuning knobs on a massively complicated musical instrument) allow inputs to create the same outputs a natural occurrence would have produced.

So, in a nutshell, technology can now mimic the behavior of systems that we humans don’t actually understand. The applications of this are boundless: We need not understand how and why language works to create a talking LLM. We need not understand climate complexity to predict weather patterns with stunning precision. In theory, we need not understand the psychological nuances of the stock market to build a neural network that predicts its behavior.

This brings us back to BCIs. We need not understand the magic behind the human brain in order to train a neural network that can replicate the brain’s functionality. One would have thought a prerequisite for creating neural implants might be a complete understanding of neuroscience. But this is no longer the case.

The technology always seemed far-fetched, primarily because we understand so little about how the human brain works. We are in the primitive stages of grasping how thoughts form, how they traverse the network of trillions of neurons we have, etc. and yet we’re expected to develop a technology that can integrate with this? That’s akin to expecting to build a rocket that can go to the moon with only a middle school understanding of physics.

The future of artificially enhanced biological brains will occur with the implantation of biologically-inspired artificial brains. The artificial parts of our intelligence will be artificial intelligence. Neural networks embedded in our neural networks…

A quick aside: If you’re enjoying this article and aren’t yet subscribed to this free newsletter, click here to get thought-provoking articles in your inbox weekly.

Mind Beyond Body

This technology is coming, faster than we expected, and without the foundational neuroscience we once thought we’d need.

In his fantastic NYT Magazine article about the topic from 2022, Ferris Jabr wrote:

“In the history of life on Earth, we have never encountered a mind without a body… All human innovations have depended on, and thus been constrained by, the body’s capacity to physically manipulate whatever tools the mind devises. If brain-computer interfaces fulfill their promise, perhaps the most profound consequence will be this: Our species could transcend those constraints, bypassing the body through a new melding of mind and machine.”

This is only one of the numerous philosophical and ethical implications that will soon face humanity as this technology becomes first available, then perfected, then ubiquitous.

In the past, mankind spent centuries exploring the significance of scientific advancements long before they took over the world. These days, it seems, we’re often forced to figure out what they all mean long after.