The lecture presents the key arguments of my forthcoming book which explores how data
and artificial intelligence abstract people into new kinds of subjects. It argues that digital
subjects are not extended selves but abstractions that operate at a distance from us,
where agency seems to be removed from our reach but which is also a condition for the
possibility of agency. Modelled and predicted, abstract subjects do not have untroubled
correspondence to actual living people. They are not us, and yet they are no one else than
us. How is it possible?

To come back to us, digital subjects engage our desire. We come to desire these
abstractions that offer us objective truths about us and the world, to be recognized or
otherwise dealt with subjectively. Thus, we inhabit abstractions through desire and as
ideals. The book–and this talk–argues that the intersections of data patterns, computational
models and scientific frameworks formulate ideals as possibilities for kinds of personalised
and yet still group subjects. Desiring the ideal subjects of AI (such as those of profiling) is
simple because we are already trained to desire abstractions such as the normal and the
best or have learned to be undone by them. The ideal then is about how abstractions are
used and how people live by them, becoming subjects.

Finally, ideal, abstract subjects get grounded by using modernity’s imaginary that
there are “real” people, “down there”, underneath the proliferation of probable subjects,
constructing the body as an anchor and arranging data worlds into one singularly possible
reality.