Page 273 - The Black Box Book

would be just as important as the reference to the physical dimension appearance, the motivation for their replacement is precisely that they
of creative work, or to the fact that creativity as a socio-cultural prac- should behave in a fundamentally different way, which is to behave like
tice has a historical dimension. However, one thing always seems to be robots in the literate sense, corresponding to the fantasies or desires of
certain in this context: AI does not have all this. It cannot achieve all this their husbands.
and never will. More importantly, the latter scenario is no longer confined to
But the possibility of distinguishing between man and machine is not the spheres of art and popular fiction. 2018, with its Duplex telephone
based on a static or stable anthropological difference. Instead, it marks assistant system, Google was able to demonstrate that machines can
a historical process that includes the possibility of profound, unforeseea- communicate with humans within certain limits without revealing their sta-
ble transformations. Accordingly, projections about the future capabilities tus as non-human beings. In view of such developments, the expectation
and properties of computers are fundamentally just as questionable as that after decades of failed attempts, computer systems will finally pass
all-too-convincing determinations of what machines will never be able to the Turing test appeared again. Indeed, it would be a milestone in the
achieve. In both respects, as history shows again and again, experts have history of information technologies, but we are not there yet, although AI
often been proven wrong. Besides, what AI can and can`t do was already technologies like GPT-3 raise expectations that this could change very
addressed by Alan Turing and indeed most of the typical projections of soon.
AI prominent in current discourses are already to be found in his writings
(Dotzler, 1989). FORMS OF AI CREATIVIT Y

POPULAR IMAGINATIONS OF AI CREATIVITY As a reminder: Back in 1950, Turing proposed to replace the ontological
question of whether computers can be intelligent or not with a test by
There is a recurring tendency in public debates to discuss AI only in means of which it should be determined whether computers can evoke
the utopian (or dystopian) mode. The computer scientist Bertram Raphael or simulate the impression of intelligence. More than 30 years ago,
(1971) once suggested to use AI as an umbrella term for problems “which Margaret Boden conceived of machine creativity similarly (1991/2004).
we do not yet know how to solve properly by computer” (p. 101). In other The machine creativity would not be concerned with determining whether
words, once computers successfully master certain tasks or a solution and to what extent the machines are creative or not, but whether they
finally emerges, they are no long considered AI. can, within certain limits, create the impression of creativity, by which she
There are several reasons for this phenomenon. For example, means producing ideas or artifacts that are new, surprising, and valuable
empirical research and development of AI technologies are strongly under (Boden, 1991/2004, p. 1). Accordingly, Boden differentiates three catego-
the spell of their cultural imaginaries. And the latter, in turn, concerns ries of creativity: combinatorial, exploratory, and transformative. These are
not only the progressive potentials of technologies but, as is well known, elaborated by Boden in relation to a “conceptual space” as a “structured
above all their negative implications. style of thought” (Boden, 1991/2004, p. 4) and discussed on the basis
How often are the real-world consequences and dangers of AI in of different software applications with reference to their performances
the near and distant future illustrated with images of HAL or by and with computers.
the Terminator? But whatever shapes our notions of intelligent machines, As examples of programs that master a combinatorial creativity,
societal fears will not be overcome by simply giving them a cute, infan- i.e. they can combine known things in unknown ways, she used, among
tile appearance or enabling them to perform amusing actions. When others, JAPE (Boden, 1991/2004, p. 3). This is a computer program
Boston Dynamics literally made its robots dance at the turn of the year for generating punning riddles, the original version of which was devel-
2020/21, this performance had obviously something uncanny about it. oped by Kim Binsted in 1993 as part of her Master of Science thesis at
The emergence of this specific feeling certainly has to do with the fact the Department of Artificial Intelligence at the University of Edinburgh.
that, on the one hand, machines resemble us increasingly in their outward The second type of so-called exploratory creativity aims at the ex-
appearance, while, on the other hand, they continue to be recognizable as ploration of a potential, which is to a certain extent already present within
machines. The latter applies for the art-project, the drawing robot Ai-Da a conceptual space but has not yet been discovered or exploited as such.
(2019) as well as for the robot Ava in the movie Ex Machina (2015). But As an example, Boden mentions the composition program Experiments in
independently of this, this sense of unease is also influenced by the fact Musical Intelligence (a.k.a. as EMI or EMMY), developed by David Cope
that media culture repeatedly confronts us with imaginations in which in the 1980s. This program was already able to analyze a composition
the difference between man and machine is not discernible. A spe- and break it down into different components, identify the style based on

cial case in this respect is the film The Stepford Wives (1975). While its patterns, and rearrange the different components into new patterns
the women ´replaced by machines´ are visually identical to their old human without reproducing anything identically.
272 273