The Collaborative AI Consortium

There is a lot of chatter around every new technology that launches, and usually it involves how it is going to replace a previous technology. But with AI the chatter is more philosophical than I have every experienced, possibly because the ‘previous technology’ it is looking to replace is humankind.

While this may seem hypey, the clue is in its name. And there are other clues in the focus of AI discussions by the tech community about how close it is getting to replicating Creativity. There’s a new Rembrandt clone, a new poem, a new piece of music created on a frequent basis. Yet, the issue which I have struggled with recently is: Why Art?

Surely there are more productive challenges that AI can be used for than putting out of business the Creative Industries who exist on a margin anyway, and provide us with so much entertainment?

At Charisma.ai, we believe that AI is a new toolkit for artists, writers and musicians. It can be used to inspire, accelerate, experiment and collaborate with, not to replace. To that end, we are proud today to announce the launch of a new group to increase the presence of the Creative Industries in the development of AI. To assist in building a technology future that is morally and creatively useful, rather than a future where literature, art, music and our creativity is absorbed without thought into technology engines with nefarious intentions (Facebook, Google, Amazon, Tencent for example).

There is a lot to do. We have taken first steps with Charisma.ai over the past few years, building a system from the ground up involving the people who will use it, pioneering moral and ethical answers along the way, with the goal to transform existing media into a new form of storytelling for those who enjoy writing and consuming immersive interactive stories.

Our approach is to explore where real worth can be generated for the Creative Industries through the use of AI. To achieve this we will adopt a simple method: ask the Creative Industries.

To keep in touch with developments on this exciting new programme, follow our Twitter account, or as always get in touch with us directly at hello @ charisma.ai

The full text of the press release and initial position paper is below:

9th February 2021 

Creative AI paper published to influence developments in artificial intelligence for creative industries

A group including representatives from The University of Oxford, Falmouth University and Royal Holloway, University of London today announced the formation of a new consortium of academics, creative industry professionals and AI specialists to influence the future of artificial intelligence developments in the Creative Arts and Humanities.

Spearheaded by a team from The University of Oxford, dbs-i and Charisma.ai, the Collaborative AI Consortium aims to rebalance the contribution of the UK’s £101.5bn creative industries to the rapid developments of artificial intelligence technologies.

At the same time, the group has published the outline article, AI and Creativity: Investigating a Human-Centred Approach. The paper highlights how current artificial intelligence developments have focused on replicating and replacing creative processes across art, writing, music and film. It argues that current AI approaches in the creative Arts and Humanities fundamentally lack an emotional connection and affect that is the foundation of creative thinking and which is at the centre of the many of the creative industries.

The focus of the the group will therefore be to explore:

  1. How we might build an artistic and human context into AI/ Algorithmic Creation 
  2. How we might draw upon Humanities research to better understand the notions of originality, ownership, emotional affect and creativity which will be at the heart of the future of a creative AI?
  3. How we could build the importance of debates around collaboration, authorship, community, inclusion and diversity into the methodologies of AI and algorithmic creation?
  4. How the creative industries and humanities could feed into a more responsible ethical and legal framework for the development of a creative AI?

The Creative AI group is comprised of representatives from Oxford University, Falmouth University, Royal Holloway University of London, Sound and Music Innovation lab, dbs-i, dBs Pro music studios, the AI-powered storytelling company Charisma.ai, Wotelsat UK, and observers from Film London and the AHRC.

Abigail Williams, Professor of English Literature and Director of Innovation Faculty of English, University of Oxford said: ‘Now is the time to link Humanities research with new technologies and work together to explore future thinking about creativity, affect and culture. This group represents exciting cross-industry perspectives, and we look forward to growing its representation in the UK and beyond through launching this paper today.’

John Matthias, Composer, Musician and Head of Research at dbs-i, said: ‘In current AI approaches to music creation, forms are typically achieved at iteratively until we are left with a piece which, perhaps sounds almost like Mozart (and has been shown to be indistinguishable from Mozart to an audience). These approaches are, however, a very long way from producing music which is anywhere near as affecting and as emotionally complex as a piece of music such as, say, The Pogues’ apparently simple song, A pair of brown eyes. This project will address this conceptual and contextual gap from centre stage.’

Guy Gadney, CEO, Charisma.ai: “It is natural that AI systems that are designed for creative purposes should have creative thinking at their core. However to date, the creative industries have been used as data sources rather than inspiration. Our goal is to bring a human-centred approach to the next evolution of artificial intelligence, to help champion diverse ideas, bring better imagination and more equitable strategies.“

Please see attached paper below for more details.

Contact:

Abigail Williams: abigail.williams@ell.ox.ac.uk 

John Matthias: john@dbs-i.org

Guy Gadney: guy@charisma.ai

----------

AI and Creativity: Investigating a Human-Centred Approach

This document outlines a Research Proposal from a consortium consisting of several UK Universities and representatives from the Creative Industries which considers a multi-disciplinary investigation into a Human-Centred approach to Artificial Intelligence and algorithmic creativity across the Creative Arts.

 

Research/ Industry Context

Many recent developments in Artificial Intelligence have been driven by technology companies who have applied it very successfully to solve advanced problems. Methodologies tend to be well defined and iterative - processes which machines have proven to carry out very well, often at phenomenal speed.

Algorithmic or AI approaches in the creative arts have been successful in replicating stylistic form (for example producing a symphony very similar in style to Mozart or a drawing in a style very similar to Picasso), but the essential inability of algorithms to comprehend a meaningful human-centred context means that moving beyond mere replication and beyond ‘holding up a mirror to ourselves’ is a big challenge for AI and its application in the Creative Arts.

Furthermore, Artificially Intelligent algorithms do not experience emotions. When we experience great art, film, literature or music, we have the experience physically. We do not convert our emotional reaction into words or text and subsequently have the experience. This experience and the recognition of the layering of all of the contexts present in the artworks happens immediately and viscerally.

This is not the way in which emotional reaction to the creative arts is currently modelled in automated approaches to artistic creation. In almost all approaches, an emotion is suggested in a word or a cluster of words and then these are converted into an artistic interpretation. Alternatively, the reverse happens; art is interpreted by a machine and converted into words which might then be interacted or adapted. In AI approaches, stylistic (rather than emotional) forms are achieved iteratively until we are left with the music which perhaps sounds almost like Mozart (and which has actually been shown to be indistinguishable from Mozart to an invited audience). These approaches are, however, a very long way from producing new creative arts which are in any way emotionally affecting and emotionally complex.

The Humanities and creative arts have a long history of thinking about how art creates affect, and how arrangements of notes, colours, or words produce particular kinds of response, and how and why they do that. These fields also have much to tell us about the relationship between form and individual creator, pattern and disruption – issues which are essential to take on board if we are to move beyond an understanding of AI-generated art as pale mimicry and pastiche. Little of this thinking has fed into the current understanding of AI and its role in creative production. This contextual and emotional gap leads to a very narrow window of aesthetic choices in the creation of art by algorithms and excludes many established and perhaps more nuanced practices such as an understanding of the importance of error and mistakes, the aesthetics of minimalism and novelty, what can be achieved and gained from the formation of artistic community and collaboration, a consideration of authorship and ownership and the inclusion of diverse opinions and voices from diverse cultures.

 

Research/ Industry Questions

This contextual Research gap has led our consortium to propose the following overarching Research Questions

1. How can we build an artistic and human context into AI/ Algorithmic Creation which leads to a production of a complex and emotionally complex aesthetic within art/ moving image, music and text?

2. How do we draw on Humanities research to better understand the notions of originality, ownership, emotional affect and creativity which will be at the heart of the future of a creative AI?

3. How can we build the importance of debates around collaboration, authorship, community, inclusion and diversity into the methodologies of AI and algorithmic creation?

4. How can the creative industries and humanities feed into a more responsible ethical and legal framework for the development of a creative AI?

Methodologies: collaborative-AI

We propose a central methodology to address these research questions, namely the creation of a new collaborative-AI -a framework which will be the emergent result of a large number of pilot projects which will place humans at the centre of the algorithmic creative process between humans and machines in many artistic domains supervised by our consortium of stakeholders from UK Universities and the Creative Industries including:

Prof. Abigail Williams

Professor of English Literature and Director of Innovation Faculty of English, University of Oxford

Guy Gadney

CEO

Charisma.ai

Dr. John Matthias

Composer, Musician and Head of Research, dbs-i

Bristol and Plymouth

……………………………………………………………………...

Jay Auborn

Musician and Producer and creative director of dBs Pro Bristol

Nuno Barreiro

Senior Lecturer in Immersive Engineering

Royal Holloway University

Dr. Sue Denham

Director of Research 

Wotelsat UK

Dr. Norah Lorway

Lecturer in Creative Music Technology

Falmouth University

Dr. Kingsley Marshall

Head of Film & Television

Falmouth University

Jordan McGarry

Head of Talent Development & Production

Film London

Annette Parry

COO

Charisma.ai

Dr. Edward Powley

Associate Professor (Games Academy)

Falmouth University

Prof. David Prior

Director of Research, Falmouth University

 

The team will initiate creative projects based in text, art, moving and still images and music including intersections between these media which will engender collaborations between human artists and AI algorithms. Many of the most innovative and exciting creative developments in the last 500 years have been the result of humans working with machines and new technologies and we will be mindful of this as we formulate our collaborative, creative AI projects.

 

Outputs and Contributions to knowledge

We expect the outputs to include 8 large-scale International and publicly showcased pilot projects which will be produced from collaborations between the artistic and academic team and/ or commissioned by the team addressing the research questions and research context in addition to multiple smaller experimental interventions. The team will also organise regular symposia and conference events inviting contributions from worldwide experts from the academic and creative/ technological worlds to address the major themes within this project which we expect to feed into edited journal issues and collaboratively edited research texts.

The new framework of collaborative-AI will be published as a set of guidelines of interaction to provoke debate within academic/ creative and tech communities worldwide.

Sign up to our newsletter!

envelope