As it appeared in venturebeat.com, by Pascal Kaufmann
May 28, 2017
There has been a lot of excitement recently about developments in brain-computer interfacing. According to the team at Facebook, pretty soon we will be able to type using our minds and hear through our skin. And according to Elon Musk, one day soon we may be able to upload and download thoughts.
Ideas such as the ability to type 100 words per minute with your mind, or learning kung fu a la The Matrix have been flung around as near-future technologies. But really these are still wild tales of science fiction. Even the neuroscience community — which rarely agrees on anything — agrees on this.
A great story
A great story about being able to reprogram our brain or hear with our skin makes gets the public and investors excited. But these stories won’t come anywhere near reality until we first crack the Brain Code. Without truly understanding the language of the brain and how billions of cells cooperate to give rise to the phenomenon we call intelligence, we won’t be able to start figuring out how to enhance it with machines.
Old and new views on AI
For all of the focus on the brain, particularly over the past 50 years, we still know surprisingly little about it. An apt analogy for the brain may be that neurons work very similarly to ant colonies. In an ant colony, every ant has a specific task: go out, find food, take a bite, leave a trail for other ants to find it. The result is one of nature’s most complex social societies and most adaptive superorganisms. From the outside, though, we don’t fully understand the full variety of roles that individual ants play in order to keep the superorganism going.
We face this same issue when trying to crack the Brain Code. Neurons, despite their quantity, only make up a very small percentage of cells in the brain. What vastly outnumbers neurons in a ratio of about 9:1 are special cells called glial cells. Scientist know very little about what glial cells actually do. We have ideas and theories but no definitive formula. Before we start building devices to plug into the brain, we must understand the full function of glial cells.
There are those who equate the brain to a computer, but that is a false analogy. Brains and intelligence did not evolve to solve algorithmic problems in an abstract virtual reality. The brain evolved with the singular goal to successfully guide an entire organism through an everchanging environment. Treating the brain as an algorithmic problem is one of the largest conceptual hurdles in moving AI forward. Only when we free ourselves from this misconception will a breakthrough in brain interfaces and AI be possible.
Musk’s idea means killing a lot of neurons
Elon Musk’s Neuralink is working on direct brain interfaces. A rudimentary form of this technology is already used to record brain activity. However, two major challenges stand in the way of Musk’s goal. Modern direct interfaces use a couple hundred tiny needles to go into the brain matter and make direct contact with brain cells. To get the level of clarity needed to achieve a real picture of how these cells work you would need hundreds of thousands (if not millions) of needles per implant. Incidentally, these implants kill the brain cells they are monitoring. The clearer you make readings of these implants, the more brain cells you kill. This is not a long-term solution and certainly not a user-friendly one.
We’ll get there eventually, but not soon
Many of the ingredients for the technology Facebook and Elon Musk are touting already exists today, but just because the ingredients are there doesn’t make you a chef. To be a chef, you need a recipe. The Brain Code is that recipe.
We will not be able to achieve any of these incredible advances in elevating brain function with machines until we’ve cracked the Brain Code. Once we understand how the brain works and how we can apply the underlying principles to artificially intelligent devices, the easy part will be applying that knowledge to the sort of advancements Mark Zuckerberg and Elon Musk are talking about. This will not only be about advancing the human body and turning us into cyborgs, it will also impact how we coexist with our AI creations such as robots, sentient machines, or even artificial humans.