Visiting with an AI Oracle or Stepping through a Looking Glass?
Image: Stable Diffusion*, generated 08.11.23 with prompt ‘Visiting with an AI Oracle or Stepping through a Looking Glass?’
Some closing thoughts, by our Editor, Gemma Milne, and The New Real’s Deputy Director, Matjaz Vidmar.
The advances in Generative AI systems have crossed an uncanny valley. They are seemingly able to conjure up a visual representation of any imaginary tableau as well as hold entirely original, unscripted conversations and perform useful editing tasks. This advancement has moved Generative AI from an exotic research field to the main-stream everyday conversation.
But what have we learned from engaging with these technologies? Are the questions we’re asking being answered, or are we simply left with more questions - ones that are really about us as humans and less about the technology itself?
Exploring the Future, with an AI Oracle
Experiencing such “intelligent” interactions showed us that perhaps a synthetic future is possible – one in which smart machines and AI enable a world of creative leisure and personal fulfilment by taking away the hard graft of manufacturing, distributing and disposing of “stuff”.
Alternatively, a more dystopian reading of these capabilities leads to erosion of human jobs and rights, and leaves a large portion of us economically, socially and politically redundant. These sort of utopian and dystopian prophecies have been expressed before, however, the present realisation of some of these systems has specifically manifested the biggest fears and hopes associated with it.
Like visiting with an (AI) oracle, the answers we are getting are not entirely in response to the questions we asked, but nonetheless, they seem to affect a foretelling of what is to come.
The limitations and concerns surrounding Generative AI are important to remember and put in context with the current hype regarding AI becoming either humanity’s salvation or its doom. Essential to the understanding of which, if any, of these futures may come to pass is the un-black-boxing of the surface-level user interfaces and the underlying algorithms and databases.
Seeing Blurry Reflections of Ourselves
So what are we seeing when we explore these technologies with those limitations and concerns in mind?
It is important to note that the technology involved is still limited in fundamental ways of “knowing”. Though Generative AI is able to build and then navigate a sophisticated model of symbolic language(s), both textual and visual, we intuitively believe that “understanding” lies elsewhere.
For starters, AI is limited to the data that it “ingested” during its “training”, and although the process is, to a degree, similar to human “learning”, its algorithmic formulaneity leaves us feeling deeply suspicious and superior to such a mechanistic recipe.
This means that while AI can generate sophisticated symbolic language models, these are not equal to conceptual knowledge models. In a sense, it is able to speak well, without knowing what the words mean.
More critically due to this explicit data-driven learning process, any novel, creative extraction of the model contents is assumed to have a strong link to the input data, hence challenging the notions of the originator's data rights and value(s).
Like stepping through the looking glass, the world behind these polished tools may seem to be strange and incomprehensible, but the reflection of our realities contained within allow us perhaps a more clear view on the challenges we face in the world outside the mathematical modelling.
A Human Exploration of a Synthetic Future
The present Edition explores these issues from a number of angles, drawing from the artistic and intellectual engagement with the new real in advance of the explosion of Generative AI into the public consciousness.
The three ever-present themes stand out:
AI is intensely political;
Technology is not necessarily ethical;
The human condition is deeply personal.
The artistic works of Jake Elwes, and Caroline Sinders and Anna Ridler, commissioned by The New Real, stand in conversation with the analytical pieces challenging the techno-moral status quo (Vallor) or problematising AI’s stewardship and cinematic embodiment (Speed). These Reflections are complemented by innovative curatorial approaches (Troiano), social-scientific interpretations (Currie and Catanzariti) and insights of new forms and responsibilities of artistic activism (Bennett). The rich contours of the critical AI arts landscape are further signposted in Interjections and Explainers covering a range of particularly burning issues from debunking myths about the AI, to mapping societal impacts of deepfakes.
This inherent complexity of navigating the new real is tackled in the Roadmap feature section, canvassing critical strategies for engaging with the Generative AI from academic, artistic, technological, curatorial, and organisational points of view. Bringing together a diverse range of voices and practitioners, we hope to champion inclusivity and diversity, whilst at the same time provide clarity and guidance. Crucially, the strategies presented are not about how to survive within a field of rapid transition, but rather they speak of opportunities for empowerment: Yes, Generative AI has brought about challenging times for creative practice, but creative practice can in turn challenge the times.
The Offering from the Arts
In this sense, if era-defining Generative AI technology is foretelling a synthetic future as perceived by today’s technological and social concerns and opportunities, the arts consistently deliver the “unexpected”, go beyond the model and the data to explore the future as it cannot be foretold.
There is productive tension between the technical advancement in apprehension of the world, and the artistic affectation of its experience. A dialogue between these dimensions allows us to realise that the path of advancement in either is neither linear nor global.
Rather, it is situated in the living bodies, our social communities, and our individual and collective expressions, which both define and defy the expectations.
We invite you to anticipate our next Edition.
–
*This image was generated as an experiment - to see if and how AI can imagine something perhaps beyond our own capabilities, using the title of this article as the prompt. It brought up questions around the style choice, the visual hints, the original works referenced, the tone, the ‘obviousness’ (or not) of the result. It also brought up debates about intellectual property, fair use, and international copyright laws. Taken alongside the explorations throughout this issue - or maybe positioning it with respect to each contribution one-by-one - what questions are you left with?