Machine intelligence

Abstract

Under certain conditions, we appear willing to see and interact with computing machines as though they exhibited intelligence, at least an intelligence of sorts. Using examples from AI and robotics research, as well as a selection of relevant art installations and anthropological fieldwork, this paper reflects on some of our interactions with the kinds of machines we seem ready to treat as intelligent. Broadly, it is suggested that ordinary, everyday ideas of intelligence are not fixed, but rather actively seen and enacted in the world. As such, intelligence does not just belong to the province of the human mind, but can emerge in quite different, unexpected forms in things. It is proposed this opens up a new set of possibilities for design and HCI; examining the ways intelligence is seen and enacted gives rise to a very different way of thinking about the intersection between human and machine, and thus promotes some radically new types of interactions with computing machines.

Diri - the actuated helium balloon

“…In evaluation we conclude, it is challenging to implement different characters and sophistications amongst devices with highly similar interfaces. One thing to keep in mind when designing with autonomy is that, under certain conditions, humans strongly impute intelligence and might ascribe qualities to objects that these in fact can't offer [37]. Our participants demonstrated this notion of imputing intelligence into our devices.…”

“…If the interaction purely consists of a device reacting to a user's input (a device fully dependent on the user -which would present a typical interface), the interface lacks intention, its own behaviour and therefore autonomy [39]. On the other hand, when the device is not reactive to the environment at all, fully independent of interaction, it is not 'in the world'; the mindless repetition of the task extracts all interaction possibilities [37]. Proactiveness has also been suggested as one property that supports autonomous behaviour [23]; if devices for example suggest content or direct interaction they shift from being purely reactive to devices which seemingly have interests and a motivation [39].…”

Tangible autonomous interfaces (TAIs)

“…Of course to this end one might develop systems, which demonstrate human or animal-like qualities. However, research has also suggested the interesting potential of developing interfaces that expose users to complex computational (intelligent) behaviour, in more machine-like ways [25]. We argue that between Tangible User Interfaces [14] and fully-formed robots there could be a class of interaction devices that exhibits increasingly autonomous and life-like behaviour.…”

At the interface of biology and computation

“…Finally, the paper's arguments are set alongside a more general set of theory-oriented perspectives emerging in HCI in which interactive systems and the material interventions enacted by them-in practice-are being recognized as integral to ways of knowing and being [22,26,29,30,32]. Although the work we present is specific to science and biology, we see our arguments to be situated in this broader intellectual dialogue.…”