'Making a Pass at a Robot' in Life on the Screen

This essay investigates computer culture and children's experience of interactive toys and 'smart machines'. It's an interesting area - connecting child development to artificial intelligence and one often thought about in very simplistic terms. To really make sense of this it's probably important to recognise that childhood itself is an invention. I'll summarise this very quickly and try to add a few ideas that might stimulate some projects before going back to the Sherry Turkle essay itself.

As I said the first principle is that the child is an invention largely of the industrial revolution, previously (and this is evidenced in historical representations - for more on this, read Phillipe Aries 'Centuries of Childhood') they were merely small adults who went out to work and were not subject to the ideology of the bourgeois family (this is where the myth of innocence is perpetuated). Neil Postman's thesis is really useful too (in 'The Disappearance of Childhood'); he argues that the child is distinguished from the adult through literacy (Freud would have it as sexuality) and that therefore the child was invented at the time of the printing press. He concludes that the child has disappeared (the end of childhood) with digital reproduction as literacy is reconfigured. You could say: that in the dialectic of nature-nurture, the child turns out to be a machine. When it comes to play, clearly the child is the plaything of the adult as well as play in general is used to condition the child for adulthood (although there is plenty of room for resistance here). Largely, toys are meant to teach acceptable adult behaviours under advanced capitalism - gender, sexuality and power roles as well as importantly to reinforce desire and consumption. Toys are part of capital's machinery and this is evident in the proliferation of new toys - from tamagotchi to interactive telly-tubbies (Zizek's essay 'Is it Possible to Traverse the Fantasy in Cyberspace?' is really good on Tamogotchi encouraging children to be murderers). Play is an interesting area in the world of interactive media and should not be dismissed as mere play (not serious work). For instance Winnacott discusses play as essential for development in terms of the transitional object - doll or blanket is an 'in-between' the child's feelings about their parents and the adult world. Winnacott used the transitional object as a metaphor with which to make interpretations during child & parent psychoanalysis.
He discovered that children in play are capable of using anything at all as successful transitional objects, including even (perhaps especially) very low technology toys such as a piece of string, a cardboard box, sand, water, earth. The child invests objects with human attributes (such as invented friends) - in this sense the teddy bear or whatever is intelligent already. Clearly most toys would pass the Turing test if the child (and probably adults too) were conducting it. Under present conditions, it might be possible to take this further and argue that if the child is the adult plaything, then it appears increasingly likely that the child is a kind of robot. It is the child that is programmable and would probably pass the Turing (or SATS) test. This isn't necessarily negative (if we apply Haraway), as the cyborg might offer a way out of the heterosexist dead-end in the context of biotechnology. A playful use of irony in all this seems essential, and the example of E-Toys comes to mind (Jago - got anything to add on this?).

Turkle claims we tend to look at children to see what we think ourselves, added to this we project onto children our anxieties about what we don't understand (look at the many moral outrages of recent times). She confronts the reductive idea that computers are merely machines and suggests that children's capacity to recognise life in inanimate objects makes the computer especially evocative to them (she cites Piaget's work on child development as a case in point). Children are often fascinated by how things work, by breaking down systems into their constituent parts, by constructing theories. Turkle describes this as 'transparency' - in her words, 'an object is transparent if it lets the way it works be seen through its physical structure' (1997: 79). She asks if it is possible to understand complex computational objects in the same way that people have tried to understand mechanical ones. For the most part, computers remain fairly opaque revealing little on the outside of what goes on inside. She claims children begin to tackle this problem through treating them as psychological objects - as if they were alive (or had a consciousness). This is where Piaget provides useful information on the connection between the concept of life and physical activity. To the young child, any activity suggests that it is alive, later through 'motion theory' making the distinction between animate and inanimate objects. Drawing on this, Turkle's past research confronted the difficulty for children in identifying whether interactive computer objects were alive or not, and deciding where the boundaries lay between people, animals and machines. She concludes that although children do not worry that inanimate objects might think and have personality, they mostly realise the machine is not alive. They are able to theorise interactivity at a sophisticated level, rejecting Piaget and other established conceptions of child development that police childhood. She characterises this as a synthesis of advanced artificial intelligence and everyday personal computing, in which technology has become 'naturalised', and the computer 'more like a demi-person' than a machine as such (1997:85). 

If the computer has developed an 'intellectual personality... an almost-physical access to the world of formal systems' (1997: 52) of code, icons, rules, then how do we make sense of this? One obvious test of the difference between humans and machines has been the Turing Test (or Imitation game); wherein an average questioner poses questions to an interlocutor whose identity is kept secret. If the questioner believes the answers are from a human when in fact it is a machine, it passes the test and the machine is said to be intelligent - it is ''real' artificial intelligence' (1997: 86). Turkle describes the criticisms of the test by Searle, who in his 'thought experiments' shifts the emphasis from 'what computers could do to focus on what they are' (1997: 87). In this way, a machine might seem intelligent but this was simply a matter of appearances and it was simply behaving as if intelligent and/or alive. She discusses the increased sophistication of these 'intelligent' machines through the example of a text-based online program called 'Julia'. She is coded for personality, can display wit, admit ignorance and even produce typing errors - all serving the appearance of authentic human characteristics. Another example would be 'Eliza' of course (this is the infamous Artificial Intelligence 'natural language communication' programme that simulates a therapist; furthermore, for a fairly crude example you could look at the chat engine of Donald Rodney: Autoicon at http://www.iniva,org/autoicon). 

Turkle further traces the 'alternative AI' (1997: 97-102), from robots with insect-like behaviour to organisms made of computer code, and the central issue of believability accounted for the most part by the desire and projection of the user. In this way, it becomes possible to account for artificiality in terms of otherness (through the language of psychology for her purposes). In this connection, when we rely increasingly on machines (and artificial agents) to do things for us, an artificial computer object is believable only in that we bring meaning to it. Or as Turkle reckons, it becomes an extension of the self (1997: 109) - not that a programme is lifelike but that we wish it to be (much the same as most human interaction in other words). But it seems too easy to say that machines and humans are more or less programmed, and she warns against a too-easy assumption that the computer is analogous to the mind (this is why I think an understanding of subjectivity is essential in all this, see the first posting on authorship).

In general, the debate over what constitutes a human or machine has shifted away from the issue of artificial intelligence to the issue of artificial life - as biotechnology takes centre stage (more on this later in Turkle's chapter 'Artificial Life as the New Frontier', in Life on the Screen). For me, the point is not to build artificial intelligence or life but to question the artificiality of its deployment in creative endeavours (or something along those lines).