Opening Remarks

Lachlan Kermode

In early April last year, we held two days of discussion to attend to the first of the lab’s three titular terms, Models, through the conceptual lens of Humanities at the Edge of the Human, the conference’s subtitle. This year, we turn to ideas relating our lab’s middle term, Scale, through the tripartite lens of how it operates across the broad and often-interlocking domains of Software, Community, and Territory.

‘Artificial Intelligence’ has always been a confusing formulation. Though it was John McCarthy who is credited with coining the term, the phrase and field that emerged from it tracks back, of course, to Alan Turing’s imitation game, which we now refer to as the Turing test. The Turing test was devised as an experimental framing for how we (humans) could plumb the depths of the distinction between computers and humans, humans and machines. It is now also a primal scene which organizes computing’s historical development, the cradle of Computer Science’s incessant drive towards the production of computational sentience. The Turing test is also the most recent origin of the myth that a credible acrobatics in the manipulation of language in specific conditions of its communication amounts to a robust philosophical definition of the hazardous notion that steers this computing research: intelligence. Turing is only the ‘most recent’ origin of this idea, for it really reaches back to at least Descartes in the seventeenth century.

One of the ongoing projects in this lab is to clarify whether there deserves to be a hard distinction between computing research in general—or A.I. research in particular—and the humanities. As has been the case in efforts that probe the boundaries of the universe, of nature, and of the psyche in centuries past, the most scientific kind of inquiry shouldn’t be hamstrung by artificial separations between disciplines or forms of knowledge. A.I. research has had a polymathic nature since its inception in the mid-twentieth century, and we would do a disservice to ourselves by over-specializing it now. This is especially true as A.I. seems to be the only place where capital still somehow sloshes over to support genuinely academic ends, rather than servicing profit or the nation-state as a means in its mimetic application in industry. That is, academic inquiry that is not rendered an effective handmaiden in capitalist or colonial conceptions of progress.

In the next two days, we will hear from scholars, researchers, and practitioners who labor towards a real kind of scientific progress. We will hear and hold space for discussion about how scale ought to be thought about in and across these three domains, software, community, and territory. Each of these concepts has a history before and beyond A.I.‘s airy and extensive discourse; but each history and concept is also challenged by developments in LLMs and neural nets to think itself again. Now that machines can write code, what becomes of the labor and status of software? Now that certain kinds of colloquial language can be produced by computers, what becomes of community? And now that LLMs have shaken up the geopolitics of computing power, restructuring it in terms of weights, biases, and the right to access certain kinds of hardware, what becomes of territory?

This conference is structured as a series of hourly sessions. Each hour will comprise one speaker, who will present for around 30 minutes, and one respondent, who will give brief remarks on the paper before fielding questions and facilitating discussion. At the end of each day, all speakers will convene in a two-hour roundtable to draw certain dimensions of the day’s discussion out further, and to give an opportunity for further questions.

The days are threaded with breaks for coffee and lunch, and we very much welcome you all to participate actively throughout the sessions, through questions, comments, and discussion over breaks.