Early this April, when researchers at Washington University in St. Louis reported that a woman with a host of electrodes temporarily positioned over the speech center of her brain was able to move a computer cursor on a screen simply by thinking but not pronouncing certain sounds, it seemed like the Singularity—the long-standing science fiction dream of melding man and machine to create a better species—might have arrived. At Brown University around the same time, scientists successfully tested a different kind of brain–computer interface (BCI) called BrainGate, which allowed a paralyzed woman to move a cursor, again just by thinking. Meanwhile, at USC, a team of biomedical engineers announced that they had successfully used carbon nanotubes to build a functioning synapse—the junction at which signals pass from one nerve cell to another—which marked the first step in their long march to construct a synthetic brain. On the same campus, Dr. Theodore Berger, who has been on his own path to make a neural prosthetic for more than three decades, has begun to implant a device into rats that bypasses a damaged hippocampus in the brain and works in its place.
The hippocampus is crucial to memory formation, and Berger’s invention holds the promise of overcoming problems related to both normal memory loss that comes from aging and pathological memory loss associated with diseases like Alzheimer’s. Similarly, the work being done at Brown and Washington University suggests the possibility of restoring mobility to those who are paralyzed and giving voice to those who have been robbed by illness or injury of the ability to communicate. If this is the Singularity, it looks not just benign but beneficent.
Michael Chorost is a man who has benefited from a brain–computer interface, though the kind of BCI implanted in his head after he went deaf in 2001, a cochlear implant, was not inserted directly into his brain, but into each of his inner ears. The result, after a lifetime of first being hard of hearing and then shut in complete auditory solitude, as he recounted in his memoir, Rebuilt: How Becoming Part Computer Made Me More Human (2005), was dramatic and life-changing. As his new, oddly jejune book, World Wide Mind: The Coming Integration of Humanity, Machines, and the Internet, makes clear, he is now a cheerleader for the rest of us getting kitted out with our own, truly personal, in-brain computers. In Chorost’s ideal world, which he lays out with the unequivocal zeal of a convert, we will all be connected directly to the Internet via a neural implant, so that the Internet “would become seamlessly part of us, as natural and simple to use as our own hands.”
The debate between repair and enhancement is long-standing in medicine (and sports, and education, and genetics), though it gets louder and more complicated as technology advances. Typically, repair, like what those Brown, USC, and Washington University research teams are aiming to do for people who have suffered stroke, spinal cord and other injuries, neurodegeneration, dementia, or mental illness, is upheld as something good and necessary and worthy. Enhancement, on the other hand—as with performance drugs and stem cell line manipulation—is either reviled as a threat to our integrity and meaning as humans or conflated with repair until the distinction becomes meaningless.1
Chorost bounces over this debate altogether. While the computer in his head was put there to fix a deficit, the fact that it is there at all is what seems to convince him that the rest of us should become cyborgs. His assumption—it would be too generous to call it an argument—is that if that worked for him, this will work for us. “My two implants make me irreversibly computational, a living example of the integration of humans and computers,” he writes. “So for me the thought of implanting something like a BlackBerry in my head is not so strange. It would not be so strange for a lot of people, I think.”
More than a quarter-century ago, a science writer named David Ritchie published a book that I’ve kept on my bookshelf as a reminder of what the post-1984 world was supposed to bring. Called The Binary Brain, it extolled “the synthesis of human and artificial intelligence” via something he called a “biochip.” “The possibilities are marvelous to contemplate,” he wrote.
You could plug into a computer’s memory banks almost as easily as you put on your shoes. Suddenly, your mind would be full of all the information stored in the computer. You could instantly make yourself an expert in anything from Spanish literature to particle physics…. With biochips to hold the data, all the information in the MIT and Harvard libraries might be stuffed into a volume no greater than that of a sandwich. All of Shakespeare in a BB-sized module…. You may see devices like this before this century ends.
“Remember,” he says gravely, “we are talking here about a technology that is just around the corner, if not here already. Biochips would lead to the development of all manner of man-machine combinations….”
Twenty-six years later, in the second decade of the new millennium, here is Chorost saying almost the same thing, and for the same reason: our brains are too limited to sufficiently apprehend the world.2 “Some human attributes like IQ appear to have risen in the twentieth century,” he writes, “but the rate of increase is much slower than technology’s. There is no Moore’s Law for human beings.” (Moore’s Law is the much-invoked thesis, now elevated to metaphor, that says that the number of components that can be placed on an integrated circuit doubles every two years.) Leaving aside the flawed equivalences—that information is knowledge and facts are intelligence—Chorost’s “transmog” dream is rooted in a naive, and common, misperception of the Internet search engine, particularly Google’s, which is how most Internet users navigate through the fourteen billion pages of the World Wide Web.
Most of us, I think it’s safe to say, do not give much thought to the algorithm that produces the results of a Google search. Ask a question, get an answer—it’s a straightforward transaction. It seems not much different from consulting an encyclopedia, or a library card catalog, or even an index in a book. Books, those other repositories of facts, information, and ideas, are the template by which we understand the Web, which is like a random, messy, ever-expanding volume of every big and little thing. A search is our way into and through the mess, and when it’s made by using Google, it’s relying on the Google algorithm, a patented and closely guarded piece of intellectual property that the company calls PageRank, composed of “500 million variables and 2 billion terms.”
Those large numbers are comforting. They suggest an impermeable defense against bias, a scientific objectivity that allows the right response to the query to bubble up from the stew of so much stuff. To an extent it’s a self-perpetuating system, since it uses popularity (the number of links) as a proxy for importance, so that the more a particular link is clicked on, the higher its PageRank, and the more likely it is to appear near the top of the search results. (This is why companies have not necessarily minded bad reviews of their products.) Chorost likens this to Hebbian learning—the notion that neurons that fire together, wire together, since
a highly ranked page will garner more page views, thus strengthening its ranking. [In this way]pages that link together “think” together. If many people visit a page over and over again, its PageRank will become so high that it effectively becomes stored in the collective human/electronic long-term memory.
Even if this turns out to be true, the process is anything but unbiased.
A Google search—which Chorost would have us doing in our own technologically modified heads—“curates” the Internet. The algorithm is, in essence, an editor, pulling up what it deems important, based on someone else’s understanding of what is important. This has spawned a whole industry of search engine optimization (SEO) consultants who game the system by reconfiguring a website’s code, content, and keywords to move it up in the rankings. Companies have also been known to pay for links in order to push themselves higher up in the rankings, a practice that Google is against and sometimes cracks down on. Even so, results rise to the top of a search query because an invisible hand is shepherding them there.
It’s not just the large number of search variables, or the intervention of marketers, that shapes the information we’re shown by bringing certain pages to our attention while others fall far enough down in the rankings to be kept out of view. As Eli Pariser documents in his chilling book The Filter Bubble: What the Internet Is Hiding from You, since December 2009, Google has aimed to contour every search to fit the profile of the person making the query. (This contouring applies to all users of Google, though it takes effect only after the user has performed several searches, so that the results can be tailored to the user’s tastes.)
The search process, in other words, has become “personalized,” which is to say that instead of being universal, it is idiosyncratic and oddly peremptory. “Most of us assume that when we google a term, we all see the same results—the ones that the company’s famous Page Rank algorithm suggests are the most authoritative based on other page’s links,” Pariser observes. With personalized search, “now you get the result that Google’s algorithm suggests is best for you in particular—and someone else may see something entirely different. In other words, there is no standard Google anymore.” It’s as if we looked up the same topic in an encyclopedia and each found different entries—but of course we would not assume they were different since we’d be consulting what we thought to be a standard reference.
Among the many insidious consequences of this individualization is that by tailoring the information you receive to the algorithm’s perception of who you are, a perception that it constructs out of fifty-seven variables, Google directs you to material that is most likely to reinforce your own worldview, ideology, and assumptions. Pariser suggests, for example, that a search for proof about climate change will turn up different results for an environmental activist than it would for an oil company executive and, one assumes, a different result for a person whom the algorithm understands to be a Democrat than for one it supposes to be a Republican. (One need not declare a party affiliation per se—the algorithm will prise this out.) In this way, the Internet, which isn’t the press, but often functions like the press by disseminating news and information, begins to cut us off from dissenting opinion and conflicting points of view, all the while seeming to be neutral and objective and unencumbered by the kind of bias inherent in, and embraced by, say, the The Weekly Standard or The Nation.
1 For instance, if glasses are reparative, is Lasik surgery too? As William Saletan wrote years ago in Slate, is it still considered reparative when a famous golfer has surgery on his nearly perfect, but not quite perfect, eyesight so he can see the ball better? See "The Beam in Your Eye," Slate, April 18, 2005. ↩
2 According to Ritchie:
There was a time not too long ago... when a mathematician could be expected to know, if not master completely, all the branches of math. Now our mathematical knowledge is expanding so fast that even an expert in mathematics...could reasonably expect to know only about 10 percent of it all, at the very most. As long as we depend on the crude input systems of sight and hearing, and the limited storage capacity of our own natural brains, that 10 percent figure is likely to keep dropping.↩
For instance, if glasses are reparative, is Lasik surgery too? As William Saletan wrote years ago in Slate, is it still considered reparative when a famous golfer has surgery on his nearly perfect, but not quite perfect, eyesight so he can see the ball better? See “The Beam in Your Eye,” Slate, April 18, 2005. ↩
According to Ritchie:
There was a time not too long ago… when a mathematician could be expected to know, if not master completely, all the branches of math. Now our mathematical knowledge is expanding so fast that even an expert in mathematics…could reasonably expect to know only about 10 percent of it all, at the very most. As long as we depend on the crude input systems of sight and hearing, and the limited storage capacity of our own natural brains, that 10 percent figure is likely to keep dropping.↩