May 18, 2004

Interview with Andy Clark

Several months ago, I conducted an interview with Indiana University professor Andy Clark. Clark is author of several books, most recently Natural-Born Cyborgs, a book that I found to be extremely stimulating.

I interviewed Dr. Clark in the course of writing the annual report for our Ten Year Forecast program, and a highly edited and compressed version of the interview appeared a couple months ago in the 2004 Ten Year Forecast. It pained me that we couldn't publish the whole thing; but space constraints prevented us from doing so.

However, there are no such constraints here; and so the interview can appear in its entirety. Here it is.

Interview with Andy Clark

Say the word "cyborgs," and the first things that usually comes to mind are science fiction characters like the Terminator or Borg. You contend that this concept of the cyborg rests on a misunderstaning of the nature of the relationship between humans and technologies. How so?

Well, it's not so much a misunderstanding as an underestimation. By focusing on these rather scary images of machine-penetrated human flesh, the pop icon Cyborg diverts attention from a deep and abiding truth about human nature. It diverts attention from what I see as the distinctive signature of our species, our ability to enter into profound and self-transforming relationships with our best tools and technologies, whether the interface proceeds via the direct wiring of flesh to silicon (the pop Cyborg route), or via the less direct (but equally effective) interfaces of touch, sight, sound or smell. Used properly, these ancient interfaces can yield human machine hybrids as dramatic as any Terminator or Borg.

What's going today on in technology that makes correctly understanding the relationship between humans and technologies important?

Two big things, at least. The first is the generation of adaptive and individualized tools and technologies. One example would be the new generation of 'software agents': packages of code that serve a specific user, perhaps searching for news-items, stocks or goods over the web, and that can learn about that user and tailor themselves to her specific needs over time. In a similar vein, imagine a PC that learnt what major functions you used most, and made them increasingly fast and efficient (at the expense of those that you don't use). In each case, the tool becomes tailored to the individual by a kind of 'mutual learning'. The second is the development of a wide variety of new ways of interfacing between user and technology. In place of the tired old low-bandwidth links of mouse and keyboard, folk at MIT media lab and elsewhere are devising ways of making information flow tangible and manipulable using skills that come naturally to human users.

One example is an 'electronic bow' for playing a digital cello. Developed by Neil Gerschenfeld and his group at MIT, the electronic bow provides a superbly sensitive, delicately nuanced, feedback-friendly means of continuously controlling the musical ebb and flow. This is an interface which has been tuned and adapted to the human bodyover centuries of use, and to which the human cellist has devoted a lifetime of study. Why throw all that away in favor of a few buttons and a mouse? The new set-up has been used to great effect by Yo-Yo Ma in a Tokyo performance. The digital media allowed the artist to create new sound combinations beyond the reach of any normal cell, while the familiar interface allowed him to explore these new possibilities with all his characteristic flair and insight.

Another great example is a US Navy innovation, a tactile flight suit that is so intuitive in use that it allows even inexperienced helicoptor pilots to perform difficult tasks such as holding the helicoptor in a stationary hover in the air. The suit works by generating bodily sensations (via safe puffs of air) inside the suit. If the craft is tilting to the right or left or forward or backward, you feel a puff-induced vibrating sensation on that side of the body. Your own responses (moving in the opposite direction so as to correct the vibrations) are monitored by the suit and control the helicoptor. The suit is so good at delivering various kinds of information in a natural and easy way that military pilots can use the suit to fly blindfold.

And of course, into this same ballpark of 'better and easier interfaces' fall all the new, rather more penetrative, technologies that allow thought control of robot arms and so on. Taken together, advances in adaptive technologies and the creation of 'transparent' user interfaces means that it will soon be harder than ever to tell where the human user stops and the rest of the world begins.

There's been a lot of press recently on Miguel Nicolelis' work constructing a neural interface between monkeys and a robotic arm. How does that work fit with your arguments about natural-born cyborgs?

In many ways, this is a prime example of the book's central theme. As you know, the neural interface allowed a monkey to control a robot arm by thought alone. Such striking images do indeed suggest that a new dawn of human-machine intimacy may be surprisingly close to hand. The point I wanted to make using the Tactile Flight suit and other examples is just that we can aspire to extreme intimacy and ease of control even without directly wiretapping the brain in this way.

Notice also that in each case (both the direct wiring and the wrap-around interface) it is the brain's fantastic ability to adapt to the new devices and opportunities that does much of the real work. It has long been known, for example, that we can rapidly adapt to the presence of inverting lenses that show you an upside-down world. After a few days, the brain learns to compensate and the image shifts upright (at least until you remove the lenses, at which point 'normal' vision (temporarily) renders yet another upside-down world!). We also learn to make fluent use of artificial limbs, and to wield an amazing variety of sports and musical instruments just as if they were parts of our own biological form.

Yet all this is ultimately rather less remarkable than it at first appears. The easiest way to see this is to consider the simple- yet critical-fact that biological bodies grow and alter during normal development. A brain that was unable to learn to control an altered body would be of little use to the baby who grows to a toddler who grows to a teenager. Even the brain that controls the final adult body must, alas, be prepared to weather nature's blows. We age, things alter, bits wither, shrink, and drop off. Brains like ours are naturally designed to be, in a very real sense, open-ended control systems. That's why bodies matter, while the details of the body do not matter so very much. Soon the body may be biological, robotic, standard, enhanced, extended, multiple, all-in-one place or spread across several physical locations. Embodiment is crucial, yet always and everywhere negotiable.

I thought one of the most interesting parts of "Natural Born Cyborgs" was your discussion of how software agents could become extensions of ourselves. What are the technical preconditions to having agents that could constantly, reliably work with us? Do you think they would reside on a server, or be carried with us in something akin to Roy Want's "personal server"?

What's important here, I think, is the potential for certain 'intimate' tools and technologies to learn and grow with the biological agent. By getting right into the heart of the developmental matrix, a software agent or other technology could come to be so sensitively co-adapted, so effortlessly consulted, and so automatically trusted, as to blur the line between such a tool and, for example, a part of the user's own brain. We don't usually talk of someone 'using' their hippocampus, or other neural structures. Rather, these structures are part of the user, part of what makes us who we are. So too, I suspect, with the best of these near-future intimate technologies. They'll be best seen as parts of the user, not things she possesses. And for this very reason, we will want to do all we can to make them constantly available and to protect them from damage. That probably means a canny combination of wearable, on-the-agent existence and well-judged off-site back-up.

What are the legal or privacy issues that you think need to be dealt with for intimate software agents? Who would own the software, and the data?

These are tough issues, and only time and hard collective experience will ultimately be our guide. But my guess is that a trained-up intimate agent should enjoy the kind of status and protection the law currently affords to the biological body. Theft or interference would be a form of personal assault.

What impact would evolved, intimate software agents have on users' memory and identity? It strikes me that being able to forget old things-- preferences, traumas, prejudices-- is essential to the growth of wisdom, or at least maturity. How would having an agent that could recall your interests and preferences from (say) 30 years ago affect that process?

A really well-tuned agent might be one that forgets as a natural by-product of new learning. Some artificial neural network technologies display a (currently hard to exploit) version of this property. Certainly, a software agent whose job is to suggest things you may want to buy will need to be sensitive to a constantly changing pattern of preferences. But any technology that uses collaborative filtering techniques to look for patterns in a way that gives extra weight to recent purchasing activity will be able to provide this.

One emerging trend in information technology is the growth of location-based services, and the intermingling of digital data and computing with physical places. Your book was primarily concerned with the impact on humans of intimate technologies; but do you see the growth of smart or aware spaces as affecting human intelligence and identity?

Definitely. I do spend one chapter of the book on the human potential, via the automatic laying of electronic tracks and trails, to profit from forms of 'swarm intelligence' (that chapter's title, 'Global Swarming', is my favorite in the book!). Wired agents, emitting signals as they move around the world, will progressively overlay the physical world with a new layer of stored information. Some of this should be stored on location, and made available (perhaps via augmented reality displays) to interested agents. To take a mundane example, I might learn that most visitors to this spot last month went on to such and such a bar or café. Or, by sharing a richer personal profile as we pass, we might become eligible ourselves for a richer, more nuanced kind of information-sharing. If I allow access to my music buying activity, I might learn which local nightclub was the next stop of most similarly-musically-inclined tourists. All this will make us vastly smarter in some ways, though it has its own dangers too. We have to balance 'following our own collectively generated crowd' with bouts of individual, semi-random exploration. If we don't, brand new nightclubs (etc) would never get their first visitors…..!

People regularly debate the impact of video games, instant messaging, and other new media on attention spans, analytical ability, and other cognitive traits. Often the issue is posed in a binary, "is it good or bad" manner. Are there more useful ways to think about the impact of media and technology on our minds and brains?

I think there are. Every technology, tool, and notation has it's own pattern of pros and cons. Roman numerals actually (though many people are surprised by this) made certain calculations easier, though in general the Arabic system wins out. Rather than thinking in terms of black and white, good and bad, we should try to spot the strengths and weaknesses, and then attempt to balance them with other tools or actions. Making a point of browsing a physical bookshop now and then is, for example, a sensible way of counteracting the tendency of collaborative filtering systems to offer us more and more of what we already buy and like. Balancing active exploration with reliance on semi-intelligent trail-sensing software is probably one of the most important new skills we will have to teach and learn.

Throughout the dot-com bubble of the 1990s, techno-pundits declared that the Internet was going to lead to the death of literature, the book, distance, the office, and just about everything else. Where did those predictions go wrong?

Things have changed, and changed dramatically. Just look at the shift from CD's to MP3's. And at the increasing number of people who really do tele-commute. But as in all such cases, it is easy to fail to appreciate that different tools and technologies really do offer different, and most often quite complementary advantages. The paper book is a case in point: a wonderful, easy, cheap, elegant technology that also has the disadvantage of locking fixed contents into a single format. Combining the use of books and online search is currently a good way to go. The very best new technologies fit in with what we already have and use, and only gradually, it at all, end up replacing it. Email, for example, is great for some things, while 'phone calls, text messaging and face-to-face contact are great for others. It is in the combinations and synergies that the real strengths lie.

Your book "Natural-Born Cyborgs" and Howard Rheingold's "Smart Mobs" talk about a lot of the same technologies, but your books go in rather different directions. What do you think of "Smart Mobs," and the whole phenomenon of using communications and computing technologies to coordinate social action?

I think it is a great book, not just because it is superbly written and researched, but also because it charts a single emerging capacity in such depth. Natural-Born Cyborgs is more diverse, and tries to weave together many different factors, such as transparent interfaces, intimate technologies, wearable computing, tangible information flows, and collaborative trail laying. Mine is also more concerned with philosophical and conceptual issues, such as possible extensions to, and transformations of, individual persons.

It seems clear, though, that these two treatments are highly complementary. Maybe Amazon should bundle us together as a special deal!

To what degree do social movements like smart mobs encourage or support the physical and cognitive changes you've studied?

As I said, I think the two go together, since the technologies overlap and the synergies are so potent. Individual augmentation and extension, and collective communication, co-ordination and trail-laying are going to feed one another in all kinds of ways, many of which we will probably not be able to predict in advance.

Do smart mob technologies suggest a future in which the line between personal identity and group identity-- between self and other-- is recast?

Well, just for fun, imagine (and here I am borrowing a story from Don Norman's super book, The Invisible Computer) that your cell-phone is built right into your jaw and ear, allowing you to communicate with others sub-vocally and to receive their communications without other folk hearing. Smart mobs would then begin to look a lot like temporary aggregate organisms (like the cellular slime mold) wouldn't they? In a way, the member of a smart mob is already using their cell-phone or other devices as a kind of sensory extension, and as a new means of coordinating action.

Going further still, now imagine opening a channel between well-chosen neural areas in the brains of several members of a football team. We know the human brain is plastic enough to learn to make use of all kinds of new signals to guide behavior, given good feedback: that's why so-called TVSS devices, that turn visual input into a pattern of tactile stimulation on the back or (more recently) the tongue are able to provoke rough quasi-visual sensations that can be used to guide object avoidance in some blind patients. By the same token, the linked brains of the football team members might learn to signal using the new channel so as to better coordinate their individual actions. In a case like that, the boundaries between the players might slip and fade, to the point where (in action, during a football game) the situation looks less like one of a well coordinated team and more like that of a single many-legged organism chasing a ball!

Looking ahead 20 years, how do you think someone born around now will differ from people of our generation?

In some ways, they'll be quite different indeed. But in the most important ways, we will still be the same old souls we always have been. We may, for example, quite literally see more than we currently do. Play very novel forms of sensory signal (such as infra-red input) into well-chosen parts of our relentlessly plastic brain, and in time that brain will probably learn to make direct and fluent use of them in controlling action. Our species future is thus as open as anyone could imagine. The human body, human sensing, and human thought are all apt for profound transformations by new forms of intimate technology.

Yet in the most important sense of all, this is nothing new, nothing frighteningly 'post-human'. Instead, it is simply the elicitation, through technology, of striking demonstrations of the plasticity that always lies at the very heart of our biological being. In the deepest and most important sense, then, I think we will remain much as we are. There will be more of some things, and less of others: more freedom and flexibility here, less freedom and flexibility there. If we are lucky, we will use our new opportunities to better understand who and what we always have been. We will become less dogmatic about bodily form, location, and about personal and sexual identity. We will see all these things as important, as fundamental, yet constantly negotiable and irrevocably technologically penetrated. We will still fall in love, we will still aspire to run faster, think harder and live better. We will still be worse than we should be and a whole lot better than we might have been.

Are there any science fiction authors you think are especially perceptive writers on the subjects you deal with in "Natural Born Cyborgs"?

I am a big fan of Greg Egan, and also of Terry Bissom. Less well-known perhaps, but very good on this whole human-machine merger thing, is Maureen McHugh (I am thinking here of China Mountain Zhang).

And of course there are the usual suspects: Neil Stephenson, William Gibson, Bruce Sterling. An older (1952) and rather neglected (perhaps because so very disturbing) treatment of the issues about identity and our relation to our parts is Limbo by Bernard Wolf. Above all, though, I guess I'd single out Warren Ellis for the great Transmetropolitan comic book series.

Posted by askpang at May 18, 2004 04:57 PM | TrackBack
Plays in: Cyborgs
Post a comment

Remember personal info?