Introduction

Images, video clips, music, and essays are being generated at jaw-dropping speed using just a few short verbal prompts.¹1 Succession-calibre melodrama is unfolding in the couloirs of a company that has put out an application with the fastest growing user-base in history.²2 Artists are challenging AI companies in court, alleging permissionless use of their works in the development of a new, powerful technology.³3 Silicon Valley tech pundits are invoking the transcendent call of godly intelligence, to which we must conform, even if it could lead to human extinction.⁴4 Opaque, seemingly influential groups with cryptic titles - EA and e/acc - are setting the terms of what AI should mean for societies.⁵5

In 2023, AI entered its main character phase on a stage where a battle is playing out for the public perception of a technology previously predominantly associated with sci-fi depictions as in The Terminator (1984) or Her (2013). The theatrics surrounding AI implicate culture as one of the key arenas in the society-wide deployment of technologies that have reached a new level of maturation, the impacts of which are yet to be understood.⁶6 The narrative wars around AI aren’t just about words and concepts. The ways in which AI is framed and which narratives will stick in the popular imagination are deeply entangled with the ways in which the technical elements of AI will interweave with broader societal dynamics.⁷7

Instead of approaching AI as a monolithic technology spearheaded by the imperatives of a handful of companies, it is more helpful to understand how present-day AI capabilities rooted in data, and model and compute components, come into being by means of a technical stack that integrates natural resources, systems, and technologies to produce the necessary hardware and software. This exploration breaks down analysis of the process into four specific chapters of focus.

Defining Public AI offers an overview of the present-day AI stack, providing key onboarding information and insights addressed to a broad range of audiences across the cultural sector and beyond. How AI is to be understood, assessed, and approached can have considerable influence over the future of regulation, governance, and ownership at each layer of the AI stack. This chapter frames this process as one of negotiating AI’s publicness. Control over different layers of the stack can have varying degrees of centralisation and distribution and will be dependent on who participates in the unglamorous work of devising and testing new organisational, technical, and legal mechanisms for balancing societal power relations in the age of AI.

While tech industry actors and governments are the obvious macro movers with levers to determine what happens next in the field. To either prevent or accelerate dominance of individual actors across the AI stack, the cultural sector - as well as the art and advanced technologies (AxAT) ecosystem more specifically - have an opportunity to be part of new coalitions, to calibrate public-facing expectations, and to demonstrate different possible directions for AI development.

Chapter 1: Organisation considers how organisations as entities - but also organisation as a process in itself - are being transformed by the logics and operations of AI, which requires large quantities of high quality training data, advanced algorithmic models and computational power. The chapter points to areas of likely high impact for cultural organisations and suggests proactive strategies. The chapter also points to the unique affordances of cultural institutions and other AxAT organisational actors, not to act solely as demo stages for new products, but as value-adding stakeholders with the agency to make consequential operational and strategic decisions in the sector.

Chapter 2: Artist narrows the aperture to the scale of artistic practices, focusing on artists (as individuals or as studios and/or collectives) who work with different elements of the AI stack as a tool or a medium.⁸8 The chapter considers the reciprocal re-shaping of the nature of artistic work with AI tools and the expectations placed on artists as producers of creative work in a context where AI tools allow users to generate endless media. The chapter considers the value that artistic skill-sets bring not just to specific strands of AI development, but also to the evolving roles and definitions of ‘art’ in society.

Chapter 3: Ecosystem points to the AxAT ecosystem as a lab for public AI. But in order for this ambitious vision to become actionable by the ecosystem and wider cultural domain, further investment into the infrastructural development of AxAT is required. The chapter points to specific areas requiring such development, from obvious hurdles relating to the technical literacy of the wider cultural sector to ongoing campaigns for embedding new metrics for AxAT. The chapter concludes with recommendations for specific AxAT and public AI experiments that build on the insights from previous chapters.

This briefing is not a definitive treatment of these matters. It should be read as a call to action, a timestamp capturing the rapid transformations underway, offering - as with all Future Art Ecosystems briefings - concepts, arguments and insights that can be integrated into individual or collective operational agendas. Critically, FAE4 emphasises challenges, but also opportunities for setting an agenda and for steering the development of public AI.