Three conferences at Oxford in the space of two weeks. I am allowing myself to be rather pleased with that one.
At the APGRD, games and new technologies; at Corpus Christi College, faith and divine grace. Now, at St Hilda’s College (with the support of APGRD and Royal Holloway, University of London), I am attempting the obvious compromise: sacred and profane, ancient and modern, Aeschylus and artificial intelligence — and what happens when AI rewrites, interprets, and rekindles Prometheus Firebringer.
βούλευμα μὲν τὸ Δῖον, Ἡφαίστου δὲ χείρ. (v. 619)
Back home, or
into the footnotes?
The Event
- Conference
- Technologies in the Theory and Practice of Greek and Roman Performance and its Reception
- Hosted by
- * University of Oxford - APGGRD;
* Royal Holloway, University of London;
* St Hildas College; - Where & when
- Where — 25 June 2026
- Official Links
-
* 26th Annual Joint Postgraduate Symposium on Ancient Performance & Reception
* University of Oxford - APGGRD
* Royal Holloway, University of London
* St Hilda’s College, Oxford
Research summary - Managed Incompletion: AI, Technē, and Classical Reception in Dorsen’s Prometheus Firebringer
What happens when AI is asked to recreate a lost ancient play? In Dorsen’s Prometheus Firebringer (2023–), a lecture-performance, GPT-generated text offers a speculative reconstruction of the lost final play in Aeschylus’ Prometheia. I argue that the work shifts the focus from recovering an ancient text to what I call managed incompletion: keeping a work open through continuous regeneration. The ancient play remains absent; Dorsen’s version must be generated anew. Neither offers a stable, definitive text.
Prometheus’ role as the bearer of technē – the arts and practical skills – provides a way to understand this use of AI. Dorsen presents AI as an analogue of the Promethean gift: a predictive technology whose power lies in opaque processes, variable outputs and the instability they introduce. Here, technology changes how we encounter the past, rather than simply helping us restore it.
The performance makes these questions tangible. Its text changes each night, inviting spectators to interpret language generated by an algorithm. Meanwhile, Dorsen’s lecture openly cites its sources, alongside the model’s opaque, statistically generated speech, which has no identifiable author. This split raises questions about who can claim authority over knowledge and who takes responsibility for what is said. Engaging with classical works becomes a matter of accountability as well as interpretation: what follows when the skilled makers of performance – the technitai – are no longer fully human? A brief comparison with AI-assisted reconstructions of Sophocles’ Ichneutae sharpens the distinction between using AI to restore ancient works and making its ongoing generation part of the performance itself.