The following is an excerpt from GigaOm publisher Byron Reese’s new book, The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity. You can purchase the book here.
The Fourth Age explores the implications of automation and AI on humanity, and has been described by Ethernet inventor and 3Com founder Bob Metcalfe as framing “the deepest questions of our time in clear language that invites the reader to make their own choices. Using 100,000 years of human history as his guide, he explores the issues around artificial general intelligence, robots, consciousness, automation, the end of work, abundance, and immortality.”
One of those deep questions of our time:
Instead of building a external general intelligence unit, would we instead augment our own brains with computing power? In this excerpt from The Fourth Age, Byron Reese considers the implications of direct interaction between computers and our brains.
Instead of building conscious computers, can we perhaps augment our brains with implanted computers? This doesn’t require us ever to crack the code of consciousness. We just take our consciousness as a given and try to add appendages to our existing intellect. This feels substantially less alien than uploading ourselves to the machine. You can imagine a prosthetic arm, for instance, that you control with your mind. In fact, you don’t really have imagine it, it already exists. Building more and more things that interact directly with the brain—say, an artificial eye—seems plausible. Eventually, could entire computers be fitted into the brain?
Elon Musk advocates a solution like this. He wants to create a neural lace for our brains, a way to directly sync our brains to the digital world. He explains:
The solution that seems the best one is to have an AI layer [added to your brain] that can work well and symbiotically with you. . . . Just as your cortex works symbiotically with your limbic system, your third digital layer could work symbiotically with you.
What Musk proposes is way beyond the brain-controlled prosthetic described at the start of this chapter. He is talking about your thoughts and memories comingling with digital ones. This would be where you think a thought like, “How long is the Nile River?” and that query is fed into Google Neuro (wirelessly, of course), and a quarter of a second later, you know the answer. If this ever happens, expect the ratings of Jeopardy! to fall off a cliff. In addition, the historian Yuval Noah Harari speculates on what else to expect:
When brains and computers can interact directly, that’s it, that’s the end of history, that’s the end of biology as we know it. Nobody has a clue what will happen once you solve this. . . . We have no way of even starting to imagine what’s happening beyond that.
There are many who say this can’t be done. Steven Pinker sums up some of the difficulties:
Brains are oatmeal-soft, oat around in skulls, react poorly to being invaded, and suffer from inflammation around foreign objects. Neurobiologists haven’t the slightest idea how to decode the billions of synapses that underlie a coherent thought, to say nothing of manipulating them.
Three breakthroughs would be needed to accomplish a meaningful merger of people and machines, and they may not be possible. First, a computer must be able to read a human thought. Second, a computer must be able to write a thought back to the brain. And third, a computer must do both of those things at speeds substantially faster than what we are presently accustomed to. If we get all three of these, then we can join with computers in a cosmically significant way.
The first one, a machine reading a human thought, is the only one we can even do a little. There are several companies working on devices, often prosthetics, that can be controlled with the mind. For instance, Johns Hopkins recently had a success creating a prosthetic hand whose individual fingers could be moved with thought. A male subject, who had his hands, was set to undergo a brain-mapping procedure for his epilepsy. The researchers built a glove with electronics in it that could buzz each finger. Then they placed a sensor over the part of the subject’s brain that controls finger movement. By buzzing each finger, they could specifically measure the exact part of the subject’s brain that corresponded to each finger. It worked! He could later move the fingers of the prosthetic with his mind. However, this would work only for his brain. For you or I to accomplish the same feat would require a similar procedure.
Another Johns Hopkins project involves making an entire artificial arm that can be controlled by the brain. Already, about a dozen of them are in active use, but again, they involve surgeries, and the limbs currently cost half a million dollars each. However, Robert Armiger, the project manager for amputee research at Johns Hopkins, said, “The long-term goal for all of this work is to have noninvasive—no extra surgeries, no extra implants—ways to control a dexterous robotic device.”
These technologies are amazing and obviously life-changing for those who need them. But even if all the bugs were worked out and the fidelity was amped way up, as a consumer product used to interface with the real world, they are of limited value compared with, say, a voice interface. It’s cool, to be sure, to be able to think “Lights on” and have them come on, but practically speaking it is only a bit better than speaking “Lights on.” And of course, we are not anywhere near being able to read a simple thought like that. Moving a finger is a distinct action from a distinct part of the brain. Thinking “Lights on” is completely different. We don’t even know how “Lights on” is coded into the brain.
But say we got all the bugs worked out, and, in addition, we learned how to write thoughts to the brain. Again, this is out in science fiction land. No one knows how a thought like, “Man, these new shoes are awesome” is encoded to the brain. Think about that. There isn’t a “these shoes are [blank]” section of the brain where you store your thoughts on each pair of shoes you own. But let’s say for a moment that we figure this out and understand it so well that we can write thoughts to the brain at the same speed and accuracy as reading something. This too is nice, but little better than what we have now. I can Google “chicken and dumpling recipe” and then read the recipe right now. There is already a mechanism for data from the eyes to be written to the brain. We mastered that eons ago. Even if the entire Internet could be accessed by my brain, that’s little better than the smartphone I already own.
However, let’s consider the third proposition, of speed. If all this could be done at fast speeds, that is something different. If I could think, “How do you speak French?” and suddenly all that data is imprinted on my mind, or is accessible by my brain at great speed, then that is something really big.
Ray Kurzweil thinks something like this will happen, that our thinking will become a hybrid of biological and nonbiological processes, and he even puts a date on it:
In the 2030s we’re going to connect directly from the neocortex to the cloud. When I need a few thousand computers, I can access that wirelessly.
It goes without saying that we don’t know if this is possible. Clearly your brain can hold the information required for proficiency in French, but can it handle it being burned in seconds or even minutes? There are some biological limits that even technology cannot expand. No matter how advanced we get, an unaided human body cannot be made that can lift a freight train. Perhaps it won’t have to be written to our brain, but our brain can access a larger, outer brain. But even then, there is a fundamental mismatch between the speed and manner in which computers and brains operate.
There is also a fourth thing, which, if possible, is beyond a “big deal.” If we were able to achieve all three of the things just discussed and in addition were able to implant a conscious computer or an AGI in our brains, or otherwise connect to such a machine, and then utilize it to augment our cognitive abilities, then, well, the question of where the human ends and the machine begins won’t really matter all that much. If we can, in fact, upgrade our reasoning ability, the very attribute that many believe makes us human, and improve it by orders of magnitude, then we would truly be superhuman. Or maybe it is better to say that something will be superhuman and that thing will own and control your body. There may no longer be a “you” in any meaningful sense.
It is hard to contemplate any of this given where we are now. The brain is a wonderful thing, but it is neither hard drive nor CPU. It is organic and analog. Turning the lights on with your brain is not just a simpler thing than learning French in three minutes, it is a completely different thing. Those who believe you will be able to learn French that way do so not because they have special knowledge about the brain that the rest of us don’t have. They believe it because they believe that minds are purely mechanistic and that technology knows no upper limits at all. If both of these propositions are true, then, well, even the sky is no longer the limit.
Despite the evident difficulty in merging computers and people, there are numerous projects under way to try to do some of the things we have just covered. The US Defense Advanced Research Projects Agency (DARPA) is working on a project whose program manager describes as attempting to “open the channel between the human brain and modern electronics” by implanting a device in the human brain that can convert brain activity into meaningful electronic signals. The agency is dedicating $62 million to the effort as part of its Neural Engineering System Design program. And it is in no way the only one working on such a project. Several other groups, both public and private, are probing the limits of what is possible.
To read more of GigaOm publisher Byron Reese’s new book, The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity, you can purchase it here.