Daily Double-Take: the Friendly "Face" of AI

Curator's Note

Talk of machine intelligence follows a familiar, adversarial script, rehashing arguments over human sense and potential algorithmic sensibility. But this misses a subtler story: how the "figure" of AI has been drawn and redrawn via mass media spectacle. The 1997 chess faceoff between Russian grandmaster Kasparov and IBM's Deep Blue was a sombre and serious "battle to the death," its studio set decorated like the living room of an intellectual sophisticate, lined with bookshelves, leather upholstered chairs and oriental rug. Deep Blue was largely out of sight--the players hunched over the chessboard, a screen on the IBM side silently delivering move instructions. While chess served its purpose in '97, this ancient and obscure logic contest, engaged in by Cold War supermen, is out of step with the task now at hand: reconfiguring the popular imaginary to accept an unthreatening, mass consumer-friendly AI. A "battle of wits" between IBM computers and favourite Jeopardy! champions provided the perfect venue--a popular, prime time, celebrity hosted, "everyone can play" trivia game--for reintroducing the public to a friendlier, workaday AI.

On Jeopardy!, Watson takes centre stage. Instead of a backroom manipulator, we're introduced to a computational child prodigy-in-training. IBM's development team exhibits wide-eyed amazement at what Watson can do, taking equal pleasure--like ultrahip parents--in his comic failures ("Crazy kid!"). Mistakes are all part of the learning process, and this kid's backend server bulk is nothing to be ashamed of, either. Other tricks help reconfigure AI's public persona, positioning him among humans, not against them. This is achieved by humanizing Watson--through proper name, voice, and an avatar that exposes and explains his computational physiology. While these humanizations are technological, they are already normalized through viewers' consumption habits and tastes; that is, "smart" technologies are already among us, so text-to-speech software, screen technology and info-visualization graphics--which make Watson who he is--are already in use, trendy or suitably "cute." The threat of machine intelligence is mitigated by making Watson a student--of Jeopardy! games past, of language and of world knowledge. Interacting with the wise and grandfatherly Alex Trebek makes Watson a harder-edged Elmo, sharing his thirst for knowledge without the cotton stuffing.

But is Watson the only one being trained here? To what extent should the viewing public be concerned about being too easily primed for a future filled with Watson-style AI--from enterprise data management and interpretation to medical diagnostics, key applications being pushed by IBM?

Comments

Wow--what a great post!  It's made me think of a hundred things, but I'll limit myself to a few for now: Firstly, Watson's transparent tutelage on Jeopardy! was something to behold, and you are spot-on in identifying how "humanizing" the whole process was, especially in terms of how we have historically imagined AIs.   Putting Watson side-by-side with human players, as well as explaining who he was and how he worked made him fairly easy to empathize with--a far cry from Kempelen's Chess-Playing Turk, whose mechanical underpinnings were shrouded in secrecy and turned out not to be mechanical at all.  Secondly, Watson as you describe him also offers a great way to reconsider what we might term his literary-aesthetic predecessors.  Consider NORAD's WOPR computer system in War Games (1983), which tries to dissuade Matthew Broderick's character from engaging in a game of Global Thermo-Nuclear war by posing the following question:  "Wouldn't you prefer a good game of chess?"  His odd mechanical voice is, at least initially, the most ethical voice in the film. Thirdly (and finally, at least for now!), don't you find it interesting that game play has historically provided the litmus test for humanity, both in the mass media and in the cultural imagination? Here I am  thinking about the out-in-the-open approach to AIs that you've identified  as a way to review Turing's "Imitation Game," which he uses to set his "Computing Machinery and Intelligence" in motion.  The "Imitation Game" is in some ways a game of hide-and-seek, in which intelligence is not determined by transparent, face-to-face or even face-to-interface encounters, but by how well players are able communicate their respective bluffs.  In fact, in both Kempelen's Chess-Playing Turk and Turing's "Imitation Game," the social, face-to-face interaction demanded by gameplay is undermined, and  I wonder, if by humanizing Watson in the way that you've described, we're acknowledging something that's been a part of this story all along, namely, that it's not just the rational, rule-based aspect of gaming that offers a useful tool to gauge humanity, but the pleasures of collaboration, social interaction, and--perhaps most important--play.  (But on the other hand, your link to IBM's "Dr. Watson" gives me the chills. Perhaps it's all a tad too dystopian still.)

It's quite remarkable how pop culture fits in to the "priming" of the public you mentioned in your piece. Though there are many literary and media caveats warning about AI, it appears those of us with more prominent concerns as part of the "masses" are made to look like conspiracy theorists--despite some of our academic credentials. The masses aren't the major problem, though--it's the scientists working to evolve technology at a faster pace than our human-evolution can match. We'll be outmoded, and rather quickly. Watson is just the beginning....

 

Good piece!

I think Lisa is onto something about the importance of collaboration, social interaction, and play being key aspects of gaming, and why we view the ability to play a game-- with all that it implies-- as a gauge of humanity. It's not just the finished product-- Watson, or the computer in War Games-- that requires humanization, but also the development process itself.

As Christine so astutely points out, the "Watson" clip shows the IBM development team taking collective pleasure at Watson's successes and failures. I think also of the scene in The Social Network when the students have a code-off, cheering each other on (and drinking beer) late into the night. Albeit with different aims, both overlay the technical aspects of the development process-- aspects that aren't easily understood by non-specialists-- with (seemingly) universal emotions and desires-- parenthood, sociability, etc. The only way we are to understand these machines and their technical bases, these scenes seem to suggest, is through the same processes we already use to make sense of the world around us.   

Thanks, all, for your input! Lisa, I completely agree that the story extends to von Kempelen's chessplayer. Thanks to this prehistory, I think there's an expectation of fraud in most any AI demo. We know von Kempelen's chess machine was rigged, and Kasparov's allegations after the Deep Blue tournament made things questionable there, too. IBM's transparency with Watson works hard to defuse this expectation, also waylaying a spectrum of fears (whether of too-fast technological development or the singularity, as you've pointed out, Rebecca).

In this sense, it's interesting to recall that the American quiz show had its own problems with fraud--nothing to do with artificial intelligence, just unethical game producers--and managed to redeem itself. AI can stand to benefit from a similar redemption, and in this, Jeopardy! is the perfect companion, as it seems to emphasize game ethics (openness in contestant tryouts, precision in answers and scoring, buzzer speeds, etc.). It is interesting that we've relied on the game setting to gauge our humanness. But insofar as games are somehow parceled off from “regular life,” I think we hope to detect something irregular there—if not always “super”human (e.g. athletics). And as social contests with defined edges, games allow us to pronounce winners, definitively. With AI, it's always been about drawing lines in the sand--marking developers' skills as much as the abilities of machine players--in this case, rewarding both for achieving the expected trickery.

I still think we want to be tricked and amazed--but this is toxic to business practice. IBM puts its most non-fraudulent face forward to disrupt the narrative we've come to expect in man-machine contests—transfering the amazement to the development process, as Lauren pointed out, its emotional, collaborative and social practice aspects. The upshot: no longer any sleight of hand--just a labour of love, a spark of genius and, (problematically when Watson is applied to non-game scenarios), the “right” answers. In this connection, what would you say about IBM’s design process and the contemporary tendency to rewrite work as creative play?

Add new comment

Log in or register to add a comment.