May 23, 2017

The Wizard Hat Solution

  



In late March, news broke that Elon Musk had embarked on a new project dedicated to linking human brains to computers. According to the Wall Street Journal, Musk had founded a company called Neuralink, which was developing a "neural lace" technology that would "allow people to communicate directly with machines without going through a physical interface." He promised on Twitter that within a week or so a long piece explaining the project in detail would be published on the popular blog, WaitButWhy.

At that point I wrote a blog post of my own, wondering how this new endeavor fit with Musk's previously expressed concerns about the existential threat artificial intelligence supposedly poses to the human species. Musk is among the few members of the techno-science elite—Bill Gates and Stephen Hawking are others—who have publicly warned that AI is developing so quickly that machines could soon become smart enough to take control of their own destinies, and ours. My blog post asked whether Neuralink represented a defense against the AI threat or a surrender to it. Along with many others, I looked forward to WaitButWhy's piece for answers.  

The post, titled "Neuralink and the Brain's Magical Future," took a bit longer than a week to appear, and when it did its length was daunting: more than 36,000 words, or 200 pages, the size of a small book. Nonetheless dozens of readers reacted with ecstatic pleasure. Comments on the WaitButWhy web site labeled it "a masterpiece" "mind-bending," "insightful," "ingenious" and "awesome."  

My reaction was considerably less positive. The tone of the piece was occasionally charming and the stick figure drawings were fun, but overall it seemed bloated, self-indulgent, and too cute by half. Its musings about the impacts and promise of technology in particular struck me as ill-informed, glib, and almost entirely lacking in critical judgment.


The author of the piece is a guy named Tim Urban, whose WaitButWhy posts on a variety of subjects have earned him a large and passionate following. He's celebrated for disproving the notion that the Internet has killed long-form writing. One of his fans is Elon Musk, who previously had reached out to Urban, inviting him to write about Musk's other ventures. Lengthy essays on Tesla and SpaceX followed. This latest piece was written with Musk's blessing and the cooperation of Neuralink's team of researchers.  
Without further ado, then, here are three reasons I find "Neuralink and the Brain's Magical Future" confirmation of a line from Proverbs: "In the multitude of words, there wanteth not sin."

1. Get to the fucking point already.
The explanation Musk had said would appear in a week or so arrived nearly eight weeks later. That wouldn't be a problem if it had been worth the wait, but it turned out Urban could have saved himself a lot of time and trouble if he'd cut the piece by two thirds. After introductory promises that what we'll be talking about here is nothing less than the destiny of humankind, there are long, long lectures on the evolution of the brain, the physiology of the brain, the development of language, and the evolution of communication technologies, consuming thousands of words and teaching us:

         a) little that any reasonably educated person didn't already know, or couldn't easily learn from Wikipedia, and

         b) nothing at all about Neuralink.

Urban is 32,000 words in before he gets round to describing what Neuralink is up to and why Musk thinks it could counter the existential threat from AI.[1] This might explain why, outside of special-interest web sites, there's been a striking lack of response to "Neuralink and the Brain's Magical Future." Commenters in more mainstream outlets, I suspect, simply didn't have the time or the patience required to digest it.

I've taken the time, which, due to a series of interruptions in my personal life (driving across the country and looking for a job, to name two), turned out to be much longer than anticipated. Let me begin, then, by answering the question I started with: Does Neuralink represent Musk's defense against the AI threat or his surrender to it?

The answer is, Both.





Urban writes that he interviewed Musk in 2015 and asked if he would ever join the effort to build super-intelligent AI. “My honest opinion is that we shouldn’t build it," Musk said then. Two years later, his views had changed. "I was trying to really sound the alarm on the AI front for quite a while," he says, "but it was clearly having no impact [laughs] so I was like, 'Oh fine, okay, then we’ll have to try to help develop it in a way that’s good.'”

Developing AI "in a way that's good," it turns out, means finding a way to merge with it, literally. Human beings must incorporate AI into our brains, Musk believes, so that we can maintain a "tight symbiosis" (his words) with AI as it develops. If we don't, we run the risk that AI will evolve in ways antithetical to our interests and eventually decide to take charge, lest we interfere with its plans. AI must not come to see us as "other," Musk says. If it does, the best we can hope for is that it will treat us nicely, as we treat our pets.[2]

So, Musk hasn't changed his mind about AI's existential threat, but he has decided that it can't be stopped, and that if we can't beat 'em, we'd better join 'em. His strategy is reminiscent of the old gangster adage, keep your friends close and your enemies closer. Like his plan to colonize Mars, it reflects a dark vision of a dystopian future. He cautions as well that we ought not waste time acting to prevent that future—the emergence of super-intelligent AI will be upon us before we know it.

Information regarding how Neuralink hopes to achieve this "tight symbiosis" with AI emerges in the later sections of Urban's piece. Basically the idea is to develop a brain implant—Urban calls it a "wizard hat"—that would serve as a wireless interface with a "cloud-based customized AI system." Musk and Urban emphasize that such an implant represents only the next logical step along a path we've long been traveling. We're already cyborgs, they say, united with our machines. The difference here is that our machines will be inside us.

This "whole-brain interface," Urban writes, would allow us "to communicate wirelessly with the cloud, with computers, and with the brains of anyone with a similar interface in their head." The flow of information between your brain and the outside world, he adds, will be so effortless, "it would feel similar to the thinking that goes on in your head today." As Musk puts it, "If we achieve tight symbiosis, AI wouldn’t be 'other'—it would be you…" 

All of this left me with the same basic misgiving I had when I heard Musk's initial Neuralink announcement: Shouldn't we be trying to become less cyborgian, rather than more?

2. Enthusiasm Unbound



Tim Urban has a bachelor's degree in political science from Harvard, but he apparently educates himself on the topics he chooses to write about. He possess a passion for study and an active curiosity, both excellent qualifications for a writer. Unfortunately, he doesn't seem especially adept at questioning his own assumptions, or especially inclined to seek out those who would question them.[3]  

In the course of his essay Urban reveals himself to be an extreme technology enthusiast. He's convinced that the development of technology is the driving force behind the emergence of what he calls "the Human Colossus," which he depicts in one of his drawings as a stick figure standing on top of the Earth. He believes the development of "newer and better" technology is the species' "core motivation." The dramatic pace and scope of technological advance today causes him to conclude that we've arrived at the hinge of history. "The Human Colossus has reached an entirely new level of power," he writes, "—the kind of power that can overthrow 3.8-billion-year eras—positioning us on the verge of multiple tipping points that will lead to unimaginable change."






Two points this argument overlooks:

1) Enthusiasts have been declaring at least since the start of the Industrial Revolution that technology is on the verge of leading us to unimaginable change. In some ways it has, but it's also true that certain fundamental characteristics of human life—birth, death, disease, family, community, inequality, conflict, work—remain.[4] The dream is that technology will create a utopia in which we'll be freed from all needs and cares. Urban doesn't promise this explicitly, but that's the implication. He's the latest enthusiast to believe that he was born at the dawn of a new millennium. It's a way of thinking you're special. Never mind that after several centuries of identical promises, we're still waiting for utopia to arrive.  

2) Urban also shares with other enthusiasts the conviction that technological advance means progress. He makes an effort to be balanced on this score. "New technology always comes along with real dangers and it always does end up harming a lot of people," he writes. "But it also always seems to help a lot more people than it harms. Advancing technology almost always proves to be a net positive." People who think that advancing technology might not be a net positive simply haven't thought about it hard enough, he says.

You have to wonder why Urban would get into this debate in the first place, and since he does get into it, why he didn't think harder about it himself. The purpose of the project he's writing about, after all, is to prevent technology from enslaving humankind. He acknowledges a number of existing problems associated with technological advance—the threat of nuclear war, automobile and airplane fatalities, the spread of fake news, cyberattacks, the enhanced ability of terrorist groups to recruit new members, and the ability of sexual predators to find new victims—but shrugs them off as trivial compared to technology's gifts. Any number of other technology-related problems might have been mentioned, some of which—global warming and the election of Donald Trump come to mind—pose threats to the future of the planet at least as plausible as a takeover by artificial intelligence.[5]  

Urban doesn't ignore the possibility that the Neuralink project, should it succeed, could present some undesirable side effects of its own. His essay includes a section titled "The scary thing about wizard hats." That he says "thing" when he means "things" is a sign he wants to get this out of the way quickly, and he does. He lists four potential scary things:

  •   trolls might mess with the computer in your brain,   
  •   the computer in your brain might crash,    
  •  the computer in your brain might be hacked by people who want to steal your thoughts, and   
  • the computer in your brain might be hacked by people who want to put thoughts into it.
Urban devotes all of 321 words to these problems before moving on to the next section, which is titled "Why the Wizard Era will be a good thing anyway even though there are a lot of dicks."

3. Trying On a Wizard Hat



Elon Musk is convinced there's no way of stopping the development of super-intelligent AI, and I suppose that's true. We like to tell ourselves that we're in control of our machines, but despite abundant concern regarding AI's disruptive potential, there's little indication its forward motion will be guided by any authority other than its inventors and investors. At one point in his essay Urban states matter-of-factly that technologies on the horizon, including AI, will drive us into a new reality "whether we like it or not."

Urban provides few specific details of the Neuralink project, which makes sense, given that it's just started. What we do know is that Musk is funding a research team in a field where a number of other teams are already working: brain-machine interfaces, or BMIs. Urban points out that BMIs have been developed that enable paraplegics to move their limbs with their thoughts, and he reviews progress in related areas such as cochlear implants, retinal implants, and full-body exoskeletons. Assuming researchers can get the support they need, Urban expects BMIs will eventually become another miraculous technology we take for granted, as we now take for granted technologies that would have seemed impossible fifty years ago.

It's a fool's game betting against the predictions of technologists and scientists; sooner or later a lot of those predictions come true. Thus I don't doubt we'll be seeing great advances in BMIs. What we can't see are the unintended consequences that will accompany those advances. In his essay Urban quotes a Harvard neuroscientist, Jeff Lichtman, as saying that if we use a mile to measure everything we need to know about the brain, at this point we've covered about three inches of that distance. Urban also quotes a neuroscientist from Stanford, Krishna Shenoy, who says that our understanding of the brain today is equivalent to humanity's grasp of world geography in the early 1500s. Urban illustrates the state of the art with this drawing:

 

This lack of understanding isn't going to stop the Neuralink team (which, by the way, consists of a group of scientists with some impressive credentials) from developing its wizard hat, or trying to. Urban quotes a team member named Flip Sabes, who says, "If it were a prerequisite to understand the brain in order to interact with the brain in a substantive way, we’d have trouble."

Presumably Sabes knows what he's talking about, but this still sounds creepy. Given how much we have to learn, any steps we take toward building a connection between artificial intelligence and the brain are steps into a vast unknown. This is not to say I'm against research into BMIs; the hope these techniques offer people with disabilities is important and exciting. Nonetheless any talk of a workable wizard hat at this point, besides being aimed at an objective that's questionable in the first place, is pure speculation. This may be another reason "Neuralink and the Brain's Magical Future" hasn't received much attention. Talk is cheap.

Urban acknowledges the uncertainty of Neuralink's prognosis with a chart that traces the hoped-for trajectory of the project's research. The path leads from "Starting Point" to "Sustainable Business Model" to "Company Innovation" to "Match That Ignites An Industry" (which is where Musk hopes Neuralink can play a decisive role) to "Industry Innovation" and eventually to "Goal," which is the wizard hat. After the wizard hat there's a box labeled "Result of the Goal." In it is a question mark. This leads to a final outcome labeled "Increased Chance of a Good Future."  




To say that the wizard hat will lead, if developed, to uncertain results is accurate. To say that it will lead to an increased chance of a good future is an article of faith. The computer scientist Bill Joy famously pointed out years ago that researchers in pursuit of breakthroughs are commonly overtaken by a state of intoxication that obscures anything other than their pursuit. "Failing to understand the consequences of our inventions while we are in the rapture of discovery and innovation seems to be a common fault of scientists and technologists," he said.




An air of intoxication hangs over Urban's essay as well. He's clearly besotted with Elon Musk and the power of technology. This limits his view, as the view of so many enthusiasts is limited. "Ever since the Human Colossus was born," he writes, "our world has had a weird property to it—it gets more magical as time goes on." Baloney. Technology may get more magical—meaning more powerful—but technology isn't the world. The world was magical long before technology came along, and there's a substantial body of opinion (among people who have given the matter considerable thought) that technology is taking it in the opposite direction. Friedrich Schiller and Max Weber, for example, argued that modern life, driven by the same habits of thought and practice that propel technological advance, was becoming progressively "disenchanted," and the impersonality and alienation that pervades contemporary society would seem to be an extension of that process. Note in this regard David Byrne's recent essay, "Eliminating the Human." Despite his frequent use of the word "magic," Urban seems unaware of, or disinterested in, these perspectives.


Speaking of magic, Urban also seems unaware that his drawings of the wizard hat are identical to the hat worn by Mickey Mouse in the Sorcerer's Apprentice segment of the animated film, Fantasia. Based on Goethe's poem, it's a cautionary tale of a would-be magician who unleashes powers he realizes, too late, he's unable to control. This is a lesson all technologists and essayists should heed. When you put on your wizard hat, watch out for marching broomsticks.









Notes:
[1] If you're not used to thinking in words, consider that, online or in print, newspapers often limit the opinion essays they run to less than 1,000 words. If Urban's essay were posted on Medium, the site's algorithm would calculate it as a 141 minute read.

[2] In one of the odder passages in Urban's piece, he says that Musk treats the rest of humankind like a pet because he believes it's incapable of planning prudently for the long term.

[3] Urban wouldn't be expected to interview people who disagreed with Musk's plans if he'd been paid by Musk to write "Neuralink and the Brain's Magical Future." Unfortunately the nature of their relationship isn't clarified.

[4] The dreamers of Silicon Valley are working on eradicating death and disease, but so far they haven't succeeded.

[5] Trump's election has caused the founders of Twitter and Facebook especially to reconsider their assumptions regarding the unalloyed social value of their technologies. To date Russia has shown no sign of reconsidering the value of its apparent contribution to Trump's victory, although that day may come.











©Doug Hill, 2017






No comments:

Post a Comment