"Idol singer": the artificial celebrity

by Djibril

William Gibson, in his novel Idoru (1996), articulated a modern cultural phenomenon by making the Japanese singer Rei Toei totally artificial, existing only in the memory banks of her parent company's computer systems. 'Idoru' is the Japanese rendering of the English word 'idol'; the word is used of pop icons, the " idol singers" who have little or no control over their own music or image but are the tools of their producers and usually disappear as quickly as they appear. Such puppets have always existed in the West too, with striking examples including Motown, Stock-Aitken-Waterman, the winners of "talent" shows, etc., but the phenomenon is apparently particularly striking in Japan.

Gibson's Rei Toei went one step further: she was not and never pretended to be a human being. Her voice, music, and videos were all synthesised, and even her creative output was software-generated. However, she was an artificial intelligence in her own right, not merely a glossy front for the creative talents of her producers. She created her own artistic material by "dreaming" music videos and the like; she had a recognisable personality and was able to make decisions, marry, and eventually (in All Tomorrow's Parties, 2000) manipulate media and science to take on a physical form.

We are obviously far from this technology, as far as the AI is concerned, but modern cinematic (and gaming) computer-generated effects almost make this sort of imitation star possible. The ability to imitate a human is of course not the most interesting feature of an artificial intelligence, but that is a different story. I find more fascinating the phenomenon whereby the public is willing to idolise a synthetic entity. Could an entirely and openly virtual star become a successful actor or celebrity, with a loyal following, longevity, range and some degree of independence? It is probably only a matter of time.

For some reason almost all the examples we have to discuss are female. This is no doubt partly the geek effect, as predominantly male computer types like to fantasise about beautiful females. There is a large measure of the objectification of women in this phenomenon too, as a virtual star can be perfect: most of the most successful and idolised women in the movies, as in music, are young and beautiful, while men in the same businesses can be of a much wider range of ages and appearances. (I would not claim there are no exceptions, of course, nor that objectified young and beautiful men do not also dominate the screens.) But it might be argued that a strong, female fantasy character can be empowering in her own right—which is not to negate the importance of the above observations.

In the 2002 Hollywood film S1m0ne, the artificial celebrity's success is measured by her ability to fool the world into thinking she is human—on screen she is of course played by human actor Rachel Roberts (although credited as "Simone"). Her attraction in the eyes of Al Pacino's character, a film director and her creator, is that unlike a human actor she has no personality, no temptation to interfere with the creative process, and therefore will perform precisely according to the director's vision of her role: she will play any part, change her appearance without complaint, perform nude scenes without charging extra fees, and so forth. Like Pygmalion, she is a male fantasy, the perfect, pliable woman. In short, she is a fraud: her appearance an amalgamation of the databank of classic actresses' faces, her movements an emulation of Pacino's own; she is designed to fool the world. (The technology required for all of this to work is of course futuristic, although—in the film—frankly unconvincing.) The twist is that even with the evidence out the world refuses to believe she is fake, but this paradox is never played out to its potential. In fact the film cops out of all the most interesting issues, such as an audience's tolerance for artificiality, and is ultimately a weak story, little better than romantic comedy at its worst.

More recently, the singer Shystie has been dubbed a "virtual celebrity" for appearing (in the form of a CGI sprite), in the car-modding computer game Juiced, for which she wrote the theme music. Whether players of this game would recognise the sprite as the singer or not, she is no more an artificial celebrity in our terms than any other computer game character, and in fact her existence in the "real" world makes Shystie significantly less artificial than Rei Toei.

Perhaps the most ambitious all-CGI movie to date was Final Fantasy: the spirits within (2001), a film that also derives from a series of successful computer games. It was rumoured at the time that the character of Aki Ross was intended to star in other films and become a virtual actress: the computer programme that generated Aki on-screen was a very sophisticated sprite, with physical contours capable of being clothed in different costumes, assuming any position, and a huge range of facial expressions and mouth movements generated on the basis of phoneme-by-phoneme shaping (her voice, on the other hand, was not synthesised but supplied in the film by the actress Ming Na). It is interesting to observe that even with all this technology, and although the film, while not photo-realistic in the strict sense, is clearly on a different level than a Dreamworks or Pixar animation, was never meant to fool an audience into thinking they were seeing real people and landscapes; within the limits of the suspension of disbelief, this illusion is maintained by even the most CGI-intensive live-action movie. All the landscapes and scenery in Sky Captain and the World of Tomorrow (2004), for example, were rendered with CGI, and although the film was given a deliberately retro, 60s TV-movie feel, the technology was largely convincing (if only the story and characters had been strong enough to carry the movie).

Of course the extremely expensive Final Fantasy was a disaster at the box office, perhaps in part because audiences were not ready for an artificial cast, but more likely, I believe, it was largely because the story and the characterisation were both very weak. The producer Sakaguchi is a game designer, and I think the company expected the amazing technological effects to be enough to carry this film. They were not. The film made production company Square Pictures such a loss that they no longer make movies as an independent body, but have been trimmed down to the status of a special effects company. It is a shame, because Final Fantasy was really not such a bad film, all things considered. But even more so, it would have been nice to see if Aki Ross could have become a celebrity in her own right. (She started out well, appearing in a swimsuit in the centre pages of a gentleman's magazine, and soon even naked on the internet.)

A more successful celebrity is the computer-game character of Lara Croft of Tomb Raider fame. In the two recent movies Lara is played by Angelina Jolie (not virtual!), and even the computer sprite was famously based on a series of real-life models: Rhona Mitra, Nell McAndrew, Lara Weller, Lucy Clarkson, and Jill de Jong; the game-sprite was voiced by Shelley Blade, Jonell Elliott, and Judith Gibbins. She has also appeared in a wide range of television commercials played by various actresses, including the Lucozade "Gone a bit Lara" series of adverts in which a whole series of women dress and act as Lara. Nevertheless it is probably true to say that more than any of these human models (even la jolie Angelina) it is the well-endowed sprite from the computer game who is memorable, who is recognised, who is the real celebrity. And she is undoubtedly completely synthetic.

We have not yet seen Lara Croft break out of the game medium in her digital form, however. The movies clearly do not count, and fan-authored animations and more or less artistic stills (including a vast range of nude Lara poses) do not constitute official publicity. She does offer evidence that a synthetic character can win an audience's loyalty, though; I am sure that the two Tomb Raider films, weak as they were, would not have attracted an audience at all if Lara was not a household name. (The game is sufficiently popular with all levels of gamers that the 'geek factor' can not account for it all. All sorts of people like to be Lara.)

Who then are the candidates for the Rei Toei-style virtual celebrity, if Aki is retired, Shystie is no more virtual than I am, and Lara is locked in a game? The biggest challenge may well be the synthetic voice, since audiences are far more forgiving of cartoon-like appearance than they are of unconvincing dialogue. Perhaps some of the most successful entrants into the Miss Digital World beauty contest (see http://www.missdigitalworld.com/ will break out into other media. The winner of the 2004 pageant, one Katty ko, is not really virtual as her physique is closely modelled on a living Chilean celebrity. One entrant however, who goes by the name of Kaya and is designed by Brazilian artist Alceu Baptistão, is already projected to be a virtual star in her own right. Not only can one find still "photographs" of Kaya on the artist's website (see http://www.vetorzero.com.br/kaya/), but he has provided animations (with voice) as well. Kaya's face is memorable, beautiful but deliberately imperfect with freckles, a snub nose and wide mouth (not the stereotypical caucasian nymph-features of most models), but she is not yet a household name, so not really a celebrity.

The social question of all our prospective virtual celebrities being female aside, there are three questions that are still to be addressed before we can truly have a digital "idoru" star of the modern media. (1) Is the technology ready to create this star's physique? We have seen near-realistic actors and computer game characters already, and the CGI in films like the recent Star Wars prequels creates characters that are meant to be entirely photo-realistic (if alien and fantastic). Animation is also just about there. Voice is another issue: so far almost all CGI characters are voiced by human actors, as the technology is still inadequate to replicate human emotional resonances in voice. (2) Is the world ready to start idolising a virtual person? The answer is probably yes, if only the character is convincing enough: we have been idolising cartoon characters from Mickey Mouse to Shrek, who is now completely digital. Once the voice is mastered, and digital film-makers start to take the story and adult audiences seriously, audiences will have no problem with this. (3) What does an artificial celebrity have to offer over a living, breathing actor with personality, imperfections and indiscretions to keep us interested? Is the only advantage to the producer, who can truly mould the star to his or her will, like Pygmalion/Simone, or does the audience gain from this in some way? It may be, of course, that the knowledge a star is artificial adds to her mystique, makes the fantasy so much more appealing; there is no limit to what you can imagine: she will not marry, have children, grow old and spoil your dreams...

There is clearly a place for the idoru in our world; it is quite possible that the truly virtual actor is already out there, but has just not made it to the big time yet. But it is nonetheless likely that no matter how artificial our entertainment is becoming, with digital enhancements to all aspects of the screen output, with overdubbed voices and models with airbrushed-out imperfections, that the human celebrity will always be with us as well.