ghosts in the m-AI-chine is both a creative-artistic and pedagogical Practice Research project currently in development. It engages in practices of so called 'digital resurrection' or 'digital necromancy' by asking AI to respond as if it were a specific performance practitioner with canonical status who has passed away. Importantly, as the project developed, the questions not only covered the practitioner's existing work but ventured into uncharted territory thus asking the AI to engage in speculative or 'creative' thinking. In doing so the project seeks:
- to serve as a critical veh-AI-cle to test the capabilities and limitations of generative AI's engagement with the field of performance by asking ChatGPT to 'impersonate' particular practitioners and artists with canonical status who have passed away, reflecting on the ethics of such a an endeavour;
- to serve as a creative-artistic veh-AI-cle, formally speaking, to further expand my experiments with the juxtaposition of image, text, and music - whilst also drawing upon my ongoing interest in the notions of musicality, presence and liveness;
- to serve as a veh-AI-cle to pedagogically explore the possibilities of using this approach in students' learning, as scaffolding and reflexive material in relation to taught sessions (i.e. live studio workshops); as well as reflecting on students' reception/views and engagement with AI-generated material in the manner proposed.
For now, the practitioners and artists chosen for ghosts in the m-AI-chine will be selected depending on what teaching opportunities present themselves [i.e. where I might be able to build-in further ghosts into the m-AI-chine during my ongoing teaching at the Department of Drama and Theatre Arts (School of English, Drama and Creative Studies)]. In time, I hope the project might also become a way for me to honour those artists and practitioners who have shaped my own practice and artistic sensibilities through something that is both a kind of personal and AI-generated panth-AI-on, as well as a kind of private yet also public AI-generated sé-AI-nce or invoc-AI-tion.
The practitioners that have become ghosts in the m-AI-chine thus far are:
* The image for the project is a self-portrait I took on Sunday 2nd February 2025 at Tate Modern's exhibition 'Solid Light' featuring the work of Anthony McCall. The specific piece featured in the self-portrait is:
Line Describing a Cone
1973
Film, 16mm, projection, digitised
30 minutes
1973
Film, 16mm, projection, digitised
30 minutes
w-AI-rds
Though these w-AI-rds were, at first, purely a playful and perhaps puerile aesthetic/stylistic choice, in the process of working on their consistent formatting (with either AI emphasised or squeezed within a word) I realised that these new and strange w-AI-rds also offered interesting opportunities to reflect: on the spectrum between sense and nonsense, and what we might discover if we allow ourselves to travel along it; on slippages of language and sense, and associations amongst and between languages; and what might be gained if we allow ourselves to be more playful and creative in our engagement with terminology. Moreoever, I realised that both these w-AI-rds and their juxtaposition with Florence Millet's music in the video below could become a way of identifying helpful cornerstones of the project's methodological approach and ethos. [Watch this space.]
[English subtitles available. Please click the 'CC' icon.]
w-AI-rds are quantum words.
Always of human origin, w-AI-rds are linguistically slippery, somet-AI-mes by mist-AI-ke and other t-AI-ms deliberately.
On the one hand, they can and often are the result of mist-AI-kes, such as typos. Whilst AI can sometimes hallucin-AI-te and generate convoluted grammar and syntactic errors, it will not usually generate a genune typo; typos are entirely human, and I am yet to encounter an instance of a t-AI-po, which currently suggests these are not possible (as they are based on a very human kind of error).
On the ther hand, w-AI-rds can be words deliberately created by humans. In these instances, they are semantically and semiotically multifaceted by design (and not des-AI-gn). They therefore resolutely refuse easy or singular definitions because they contain within themselves a number of different meanings, and sometimes slip between different languages, occasionally even in contradictory ways.
As new terms, a w-AI-rd [singular] and w-AI-rds [plural] seek to evoke both the contradictory associations of a human typo on the one hand, and an AI-generated hallucination or mistake on the other - since the terms are made of by taking the words 'word/words' and forcefully inserting the diphthong AI into them (a diphthong is the sound formed by the combination of two vowels in a single syllable).
In a sense, w-AI-rds are like the infamously difficult to pin down Indian head bobble [a gesture that can almost indistinctly mean: 'yes' or 'maybe'; be a gesture of respect; a recognition of understanding, agreeing, or acknowledging - to varying degree; or even just a way of avoiding directly saying 'no'].
On the one hand, they can and often are the result of mist-AI-kes, such as typos. Whilst AI can sometimes hallucin-AI-te and generate convoluted grammar and syntactic errors, it will not usually generate a genune typo; typos are entirely human, and I am yet to encounter an instance of a t-AI-po, which currently suggests these are not possible (as they are based on a very human kind of error).
On the ther hand, w-AI-rds can be words deliberately created by humans. In these instances, they are semantically and semiotically multifaceted by design (and not des-AI-gn). They therefore resolutely refuse easy or singular definitions because they contain within themselves a number of different meanings, and sometimes slip between different languages, occasionally even in contradictory ways.
As new terms, a w-AI-rd [singular] and w-AI-rds [plural] seek to evoke both the contradictory associations of a human typo on the one hand, and an AI-generated hallucination or mistake on the other - since the terms are made of by taking the words 'word/words' and forcefully inserting the diphthong AI into them (a diphthong is the sound formed by the combination of two vowels in a single syllable).
In a sense, w-AI-rds are like the infamously difficult to pin down Indian head bobble [a gesture that can almost indistinctly mean: 'yes' or 'maybe'; be a gesture of respect; a recognition of understanding, agreeing, or acknowledging - to varying degree; or even just a way of avoiding directly saying 'no'].
In ways that I need to yet articulate, this could be seen as a sister project to: