Nicola Osborne, The University of Edinburgh, in consultation with Ewa Luger, John Vines, Burkhard Schafer, Drew Hemment, Chris Speed, Shannon Vallor, Melissa Terras, Michael Rovatsos, Caroline Parkinson
Recent advances in artificial intelligence and machine learning have made these technologies more accessible to creative businesses and individual artists. The creative and commercial capabilities arising from this would have been unimaginable only a few years ago. We look towards the role of Creative AI in the post-Covid-19 creative industries, as calls to establish a Centre for Creative AI are gaining momentum.
Why does it matter?
The creative industries are already exploring innovative routes to Covid-19 recovery, including new business models, experiences and forms of creative production emerging from the need to survive, invent, and respond to changing demands during pandemic restrictions. The unprecedented adoption and experimentation with new tools and technologies over the last 18 months have further accelerated interest in the practical and playful possibilities of Creative AI. Prior to Covid-19 the creative industries were one of the fastest growing parts of the UK economy, with digital technology and ambitious R&D – including AI – central to its future. However, the creative industries never stand alone and will also play a key role in helping society emerge from this challenging period.
Building inclusive AI
Computer Science as a whole, and AI technologies in particular, have been dominated by a specific masculine view of the world that brings with it a Western hegemonic legacy. In contrast, the breadth of creative industries embraces globalism, pluralism and a multiplicity of voices including Feminist and Queer perspectives, intersectionality, BAME voices and the raising of other minority and community interests through creative forms of expression – all currently absent from mainstream AI developments. One of the core benefits of a centre for Creative AI is the potential to move beyond the myopic determinism that has characterised AI developments to date and instead open the door to different perspectives and markets that speak to wider communities and reflect more complex social narratives. One need only look to the possibilities when activists show the world through a new lens (e. g. Joy Buolamwini’s Coded Gaze work) to understand the potential of creative industries as a catalyst for change. Similarly, the arts and new modes of AI-supported storytelling can help to surface ethical issues and/or explore social implications of AI in meaningful and impactful ways. For example, Kate Crawford and Trever Paglen’s work (Training Humans and ImageNet Roulette) exposes algorithmic errors and ideological assumptions, which would otherwise remain concealed. Hence, designing AI as a creative process has the potential to bring substantial benefits in exploring inclusive and criticality informed adoption of AI within and beyond the creative industries.
The creative industries, as innovators and originators of new ideas and IP, and through their engagement with wider industries and society, provide critical perspectives on the potential of new technologies for responsible and ethical R&D. A new Centre for Creative AI, bringing together the UK’s expertise in this area, could be an essential gateway to R&D in AI for the creative industries, enabling access to training and skills development. Such a Centre will allow creative research methods and practices to contribute to the development and application of new data-driven technologies and place design methods alongside centres of computing and data science.
A Vision for a Creative AI Centre
What might a potential centre for Creative AI do? Here we suggest five areas of work which we believe would form a strong blend of research and industry orientated work for Creative AI to thrive.
Landscape mapping and establishing foundations
The UK has the third highest level of research in AI in the world which gives a Centre for Creative AI huge potential for collaboration and opportunities for creative research with mutual benefit for both the creative industries and other areas. One of the first steps for a new Centre in establishing its purpose, identity and role will be to locate and connect its proposed work in the wider research (e.g. The Alan Turing Institute) and policy (e.g. UK Government Build Back Better and Scotland’s AI Strategy).
Further, to have real meaningful impact across and beyond the creative industries, the Centre for Creative AI will need to be interdisciplinary at its core, bringing together deep-rooted technical expertise in specific areas of AI alongside deep connectivity with the creative industries, broadly framed. In addition to combining design and creative researchers, and experts in computer science and informatics, it will also be crucial to connect to data and technology-informed expertise across areas including moral philosophy, business, and law, in support of an inclusive and ethical approach that generates new industry-relevant knowledge.
Interdisciplinarity is key to making the most of the potential for AI in the creative industries, but collaboration between AI and the creative sector also hugely benefits other sectors looking to make creative and novel uses of AI. Creative work spans the economy, instrumental in designing the products, services and experiences we encounter every day across banking, retail, tourism, public services, etc. Similarly, there are new value propositions for the creative industries to shape how industries explore and adopt AI in creative, humane, and critically informed ways.
Expressing and developing value propositions
AI is already central to the way creative content is produced and consumed. Music and video services use AI to recommend creative content to users. In advertising, businesses are using AI in analysis and campaigning. Increasingly, creators are using AI in their workflow. Many artists today actively choose to use Generative Adversarial Networks (GANs) as their medium, in the same way traditional or new media artists may work with paint on canvas, film, or digital video. AI is ripe for exploration and experimentation in the same way that modernism has always explored the “unique and proper area of competence” of art, investigating “all that was unique in the nature of its medium” (Clement Greenberg, 1960).
AI is not just about adopting a suit of technologies or tools, AI can be intrinsically linked to business models enabling more radically disruptive creative ideas and businesses, with the potential for significant benefits to industry and society. Through exploring and creatively challenging the potential futures of AI, Creative AI as a field not only has the potential to benefit the creative industries, but also to stimulate foundational R&D in AI with impact across a wide range of industries.
Innovation, boundary-crossing, and showcasing
The new Centre for Creative AI will be a crucible for new ideas, supporting a diverse array of creative innovation with AI, including new forms of creative expression, new experiences, and the kinds of automation and streamlining of processes that are still very manual.
Whether they are creative industries practitioners, or researchers new to Creative AI (or AI in general), there is a need to understand the potential, the relevance and some of the (challenging) realities of working with AI to be able to dig deeper towards new R&D. The new Centre will need to nurture an inclusive space for experimentation and collaboration through engagement including showcasing key work, events such as seminars and workshops, and perhaps even online marketplaces for commercial work. The Centre will expose industry and research stakeholders to inspiring creative work that pushes technological and societal boundaries (e.g. Martin Disley’s Deepfake Doppelganger for Zoom Obscura; Jake Elwes’ The Zizi Show, part of The New Real), and work that showcases the kinds of automation and streamlining of creative industries processes, which can spark excitement, new engagement with AI, and ultimately new collaborations (e.g. Viapontica’s automated image cropping system for The List; Black Goblin Audio’s tools for creating new soundscapes with procedural audio) .
Co-producing rigorous processes, methods, and skills
The centre for Creative AI will inevitably be working with research and industry stakeholders at different levels of expertise in varying levels of R&D – from cutting edge early research, to at/near market-ready technologies for adoption. A key focus will be to co-produce rigorous processes and approaches across research and industry, establishing methods and skills around Creative AI.
The ecosystem of technologies and methodologies which we know as “AI” allows computers to perform tasks that would otherwise require human intelligence and/or operate at a scale not realistically addressable by human efforts. By their nature these methods can be hard (sometimes impossible) to examine as they adapt and change in response to the data – a feature that enables enormous positive possibilities to find patterns or novel ways to problem solve (for example in improved detection rate in lung cancer screening). However, these methods can also raise societal, moral, power and influence issues, particularly if/when the data (on which systems are trained with) reflect existing biases, inequalities, and societal problems (e.g. Microsoft’s “Tay” AI chatbot, Google AI image labelling tools, AI and Automated Decision Systems in law enforcement). The established history of participatory methods within design further enables the data-sciences to meaningfully engage with publics, organisations, and creative companies to identify and navigate the application of AI. By taking a more holistic view of AI the Centre for Creative AI should be establishing robust methods and skills that encourage best practices and an inclusive, critical approach to the use of AI.
The ethical implications of Creative AI will be a key consideration for the new Centre to explore and establish best practices around. Whilst there are many researching ethical implications of AI, the focus on doing so for the creative industries, and in an applied, practical context is critical for meaningful adoption (e.g. the Creative Informatics Ethics Guidance). Further, the Centre will need to consider how best to fill the skills gap in graduates from creative disciplines entering AI-enabled industries and sectors.
There are significant emerging areas, that firmly straddle both research and industry, where real engagement and co-production approaches will be required to develop robust methods and guidance. One area of particular concern is the IP implications of work co-created with an AI, which has legal, creative, moral, and financial impacts and is still underexplored even as AI has entered the production processes in music, film, etc.
The Centre should develop and make available AI-enabled creative methodologies (ie. building on Philips’ work on data-enabled design) for the benefit of both research and industry adoption. This will ensure a theoretical depth – from both technical and creative perspectives – to support impactful engagement and showcasing of approaches to AI.
A hub for support, connection, and collaboration
For many in the creative industries the possibilities of AI are tantalising, but AI is also an intimidating area to approach. It can be extremely difficult to identify appropriate expertise, whether within academia or in the marketplace, particularly when AI and data skills are in such significant demand. A fundamental part of creating a vibrant Centre for Creative AI, and a community of collaborators, will be to create an environment that develops skills and confidence both for industries and for other researchers engaging with AI and Creative AI for the first time.
The Centre will need a strong programme of introductory activities, training, and scaffolding mechanisms, including expert advice, to develop deeper understandings of AI, actively pursue diversity, and broker rich connections towards new research and industry R&D.
The greatest benefits for Creative AI will come from working on an interdisciplinary basis but deep collaboration across divergent areas of research and industry requires robust translation across terminologies, working practices, and expectations. The Centre will therefore also need to act as a site of translation between different disciplines and contexts, building understanding and relationships through expertise, funding, and resourcing to develop a robust network for ongoing R&D. In addition, the Centre would showcase the increasing wealth of use of AI in the UK’s world-leading creative industries, demonstrating that it is a nexus worthy of business investment and development, and providing the on-ramp for more individuals and companies in the creative industries to embrace and engage with AI’s possibilities.
What next?
It is time for a joined-up approach to AI in the creative industries. We’ve proposed some ideas about how this might be manifested in a new Centre for Creative AI, and welcome the prospect of this going forward in the near future.
May 2021