Affect and Artificial Intelligence
Affect and Artificial Intelligence
by Elizabeth A. Wilson
University of Washington Press, Seattle, WA, 2010
In Vivo The Cultural Mediations of Biomedical Science-series
200 pp., illus. 6 b/w, with notes, biblio., and index. Trade, $60.00; paper, $25.00
ISBN: 780295990514; ISBN: 978-0-295-99047-7.
Reviewed by Jussi Parikka
Winchester School of Art/University of Southampton
Giggling. That is something that you would not expect to be emphasized in a book about early artificial intelligence and cultural theory. Instead of just wanting to go through the in itself depressive homophobic culture that surrounded Alan Turing and probably contributed to his suicide, Elizabeth A. Wilson wants to paint a different kind of a picture of the intertwining of affect and research into cultures of computerized rationality. Hence, Turing is not seen only as the unfortunate victim of state persecution and enforced chemical castration, but she wants to emphasize the overflowing positive affect-worlds that were intimately linked with Turing’s analytical questioning. The “positive affects”, such as Turing’s giggling and delight, are carried over to the analysis that intertwines theory with historical material, insights from his personal life, and his work – and succeeds in this really difficult genre of cultural analysis really well. Hence, the often disembodied – and also what critics have persistently labeled as narrowly defined boundaries – of intelligence that the early research into AI of the 1950s and 1960s suggested was actually embedded in a complex circulation of affects, motivations, desires, emphases, and investments. Wilson’s book on affective worlds is not just a critique of AI for neglecting such drives and affective tendencies – but demonstrates through archival work and theoretical insights that we can do more as cultural theorists.
Wilson starts the book with an epigraph from Bruno Latour and the insistence to think of critique as multiplication. Critical theorists should not be content to stay on the paranoid mode of criticizing of what went wrong – for instance the seeming lack of embodiment in such AI discourses, the phallocentric rationality, and homosocial gender bias – but how to use material affirmatively (but no less critically) to come up with novel ideas –that is, something more. As such, some of Wilson’s positions, and underlying methodological insistence reminds us not only of Latour but also of such material feminists as Karen Barad and Rosi Braidotti.
Wilson’s readings of the affective registers of people such as Turing or for instance Walter Pitts (of the 1940s fame of the McCullough-Pitts model of neurons that was the first to ground a scientific link between the wet brain and the binary logical computer.) What is intriguing are her archival take and the desire to address the relations between affect and computer-based discourse of artificial intelligence through the early phase of AI. Instead of going the more obvious route – of claiming that, whereas early AI neglected emotions and affects and proposed a narrow view of what intelligence is, and what was later corrected with the more embodied, relational and dynamic models of learning robotics – Wilson wants to point out that affect was already there. Hence, her work is perhaps psycho-biographical, but even more so about the environments of creation in which relations between people, mathematical theories, and engineering of such machines was completely filled with various affective registers. Wilson emphasizes this point at the beginning her archival work but that could be even more visible and richer in the actual analysis (as I am sure she did a lot of groundwork with materials).
Wilson never distinguishes between affect, emotion, and feeling, which might lead to some questions. At times affect means more or less emotions, but at the same time she does hint towards a richer, relational notion of affect arising of relations – even physical relations – between people and things. It also elaborates affect as an affordance – as more than a categorical “drive”, and as an intensification of our engagement with the world. As such, the book is a great intervention into the continuous debates concerning our relations with machines – most recently by such figures as Jaron Lanier and Sherry Turkle in rather pessimistic tones. What Wilson is saying with her elaborated, theoretically refined, and exciting take is that we need more introjection, less projection. We are already intermingled with our machines, with object-relations, and in affective circuits that include not only people but machines too. Bruno Bettelheim’s case study of the “robotic boy Joey” analyzed by Wilson (29-30) is a case in point.
In addition to the chosen case studies, for instance the work of “machine-machine” relations of “affect” – when affect is understood as triggering relationality – would have been a good addition (I am thinking about the work of W. Grey Walter and his robotic tortoises). In addition, when she uses affect as a way to think through “intersubjectivity” (27) – instead of for instance seeing it as the pre-individual social in the manner as does Brian Massumi – it would have been interesting to elaborate the point a bit. Is affect only between subjects, or constitutive of the subjectivity itself, already partly swallowed in other people’s and things’ world (as she otherwise seems to suggest)?
In any case, just like her previous book Psychosomatic, this is a great read and engages the reader in a kind of a “draw the lines between the dots” kind of a practice: The book has a lot of implicit links to theoretical discussions that are going on currently and Affect & Artificial Intelligence offers a nicely grounded perspective to many of these.