Head of Research, FS AI Enterprise Hub, Capgemini
I work on the layer between an institution's data and the AI agents about to act on it: what an agent may rely on, who may let it act and under what mandate, what is checked before the act, and what record is left behind.
Behind that is an older interest: intelligence, at every level where it is found and in the spaces between. A cell, a brain, an organisation, a model, an agent: each embodies it at its own level, physical, biological, digital. And much of it is between the units, as music is between the notes and an ant colony thinks in the traces its ants leave for each other. So the questions here run from a payment an agent releases this afternoon to what it means to be held, bound, or predicted by a machine.
What I am busy with, and the essays and records behind each question.
Read the questions →The newsletter in its two sections, AI and the human and governed intelligence, and the earlier articles.
Read the essays →The body of work on governedintelligence.eu: the papers, the reference architecture, the control plane, the manifestos.
Open the work →We are only seeking Man. We have no need of other worlds. We need mirrors.
Stanisław Lem, Solaris, 1961. We have always imagined intelligence as a version of ourselves, and we recognise only the kind that looks like us. Even modelling a machine on our own brain is anthropocentrism of a high order: the mirror again. The questions that follow from that.
An essay on how intelligence depends on the organisation of a system, drawing on ant colonies, brains, and groups of AI agents.
Why an approved specification may still contain assumptions that need separate checks before an AI system acts on them.
What an institution needs to establish before allowing an AI system to act, even when its output is good enough for the task.
Why a cell, a child, and a company all stop trying when nothing comes back.
Three more records went up the same day. The research page lists them.