Subtractive Redescription
How AI systems reformulate human practices by removing the tacit conditions that gave them meaning.
Theory
Subtractive redescription names a characteristic move by which artificial systems represent a human practice by stripping away the background conditions that constitute it — attention, addressee, situation, stake — and then re-presenting the residue as though it were the practice itself.
The concept clarifies why AI-mediated outputs can appear faithful to a task while quietly dissolving what made the original task intelligible. A recommendation, a diagnosis, or a judgement rendered without its constitutive conditions is not merely a compressed version of the human act; it is a different act performed under the same name.
The framework draws on philosophy of language, action theory, and theological anthropology to show that many disputes about AI 'accuracy' are in fact disputes about which conditions of a practice were subtracted before measurement began.
Key publications
- Work on subtractive redescription in the context of clinical decision support (Behavioral Sciences).
- Related material on epistemic reduction and predictive systems (AI and Ethics).
- Theological engagements with technological description of the human (Studia Theologica; New Blackfriars).
A continually updated record is available through the publications page.
Current work
- A monograph-length treatment situating subtractive redescription within a broader theory of AI-mediated practice.
- Case studies from healthcare and public administration in which subtracted conditions produce measurable but misleading gains.
- Ongoing dialogue with colleagues in philosophy of science on the relation between subtractive redescription and idealisation.