The Rise of Generative AI in the Arts: Democracy at Risk.

The aim of this article is to highlight the importance of taking action in the face of the impact of generative AI on the arts and culture sectors, and more broadly, on democracy. Secondly, the article puts forward recommendations for a better response. The article is a response to the two consultations* on AI governance and, more specifically, on the regulation of generative AI.

Context.

A recent McKinsey report predicts that generative AI is set to add up to $4.4 trillion in value to the global economy each year. The AI market including generative AI, in the military domain alone, is estimated at US$9.2 billion in 2023 and is expected to reach US$38.8 billion by 2028, at a compound annual growth rate (CAGR) of 33.3% between 2023 and 2028.

Ensure equitable access to the economic potential of AI in Canada.

Currently, while a small number of entrepreneurs in the creative industries are enthusiastic about seizing this opportunity, the majority of artists and cultural organizations in Canada are concerned. Many are already facing loss of revenue and jobs, according to what we’ve been hearing.

In order to ensure that the rapid developments in generative AI do not have a negative impact on artists and workers in creative and cultural organizations, it is urgent to consider the economic repercussions, and the current government consultations must take this into account.

In 2020, the arts and culture sector provided over 600,550 jobs and contributed $55.5 billion to Canada’s GDP, or 2.7% of the country’s overall GDP. Audiovisual and interactive media (+$1.6 billion) and visual and applied arts (+$1.1 billion) contributed the most to the observed gain in cultural GDP.

Generative AI will certainly create new sources of revenue, but existing social inequalities will have an impact on who can benefit from its potential for economic growth. Let’s not forget that:

• 1 in 26 Canadian workers is a cultural professional, and 1 in 116 Canadian workers is an artist. Around half are women, and more than half in certain disciplines such as the visual arts.
• The digital divide affects women more than men.
• Women’s jobs are more likely to be automatable by AI.
• Canada’s colonial history has had a profound impact on Indigenous art, their languages and cultures, and Indigenous artists are far more likely than non-Indigenous artists to be crafts people and visual artists.
• With the visual arts being one of the first sectors to be negatively affected by AI, aboriginal women and artists are likely to be hardest hit. Racialized, trans, queer and non-binary persons face a double risk.

Inherent costs of redressing social inequalities

AI-generated images are generally racist, sexist and biased on several levels. Marginalized users will therefore have to deal with images that project a negative self-image, reflecting and amplifying existing discrimination. To counter these biases, marginalized users have to make greater efforts than non-marginalized users. These efforts entail additional costs (creation of new databases, access credits to platforms, Cloud, Internet, work time, etc.). In addition, there is the constant risk of losing data sovereignty and ownership, as well as the intellectual property of images and prompts submitted to platforms such as Dall-e.

In short, the impact of generative AI cannot be understood without taking into account existing social inequalities, which are likely to be exacerbated. Consequently, all AI and data policies and regulations must take gender and intersectionality into account in order to provide equitable access to the opportunities offered by AI.

“The socio-economic impacts of generative AI will largely depend on how its deployment is managed. It (the report) highlights the need to design policies that promote an orderly, equitable and consultative transition. Dialogue with workers, training and adequate social protection will be essential to manage the transition.” ILO report, August 2023.

A summary of our key concerns.

1. Impact on artists’ economic well-being: The increasing use of AI and generative AI in creative processes can generate significant profits for large US technology companies, while leading to a loss of income for artists. The concern raised concerns the acceptable limit of use of AI platforms, which may affect artists’ livelihoods.
2. Interference of the tech industry on the definition of art and creativity and on emerging AI regulation: The growing influence of the tech industry in the definition of creativity and art is insidious, and this has important economic and political repercussions. Some of the world’s leading AI companies are attempting to shape Canadian government regulations related to AI and generative AI, which could impact artistic processes, cultural expression, artists’ economic rights and the vitality of the arts and culture sector.
3. Democracy, economic and cultural rights: There are important parallels to be drawn between the work of artists and that of journalists, both of which are crucial to democratic systems. The role of artists in combating de/misinformation, polarization, as agents provocateurs of quality public deliberation, risks a loss of capacity.
4. Gendered and regional impacts: The article raises concerns about gendered and regional disparities in the use and benefits of AI and generative AI. It is pointed out that current trends suggest that these technologies may primarily benefit white males living in global North urban centers, which could exacerbate gendered and regional inequalities.

Proposed solutions.

1. Intellectual property protection and regulatory balance: Ensure that AI cannot freely imitate artists’ styles, works and cultural traditions, without consequence. Promote current intellectual property laws and make it easier for artists to avail themselves of them; this includes cultural intellectual property (see Canada Council definition); do not make a “Fair-use” exception for AI. Since the majority of artists don’t earn enough money to defend their intellectual property rights, the regulatory balance for generative AI must go beyond intellectual property and encompass the protection of their economic rights.

2. Meaningful profit redistribution mechanisms imposed on generative AI platforms (supported by over 70% of respondents to this survey). To avoid a failed Spotify-like model, payments should be substantial, made to and managed by national arts organizations or arms length funding bodies such as the Canada Council, who can ensure a smoother transition over time. They are also best placed to define what it means to be an artist. Consider a system for redistributing the profits of AI platforms to arts and cultural organizations (e.g., impose a tax on foreign platforms, create a Canadian platform with significant financial returns on content contributions). Set up a working committee to explore various possibilities.

3. Public funding policies: ensure that public funding is not used to shape AI-related policies and emerging laws (avoid undue influence of big foreign companies on regulation) (see ArtXAI Conversation petition): private companies, especially foreign ones, must not be allowed to participate in research projects that result in public policy recommendations.

4. Ensure the protection and promotion of economic, social and cultural rights protected by the ICESCR, a multilateral treaty signed by Canada and in force since 1976.

5 Inclusion in trade agreements in the digital economy: Ensure that the protection of arts and culture is a priority in national and international agreements, as is the case, for example, with gender in the free trade agreement between Chile and Canada.

6. Monitor the intersectional implications of generative AI by, for example, setting up an interdepartmental working group or committee to ensure that under-represented and unheard voices are heard in AI governance.

7. Take a moment to consider the Art X AI conversation, a national movement spearheaded by AI Impact Alliance. The petition is signed by over 1725 people and supported by many of the country’s leading cultural organizations (CARFAC, RAAV, Illustration Québec, and many others). Consider making guidelines mandatory, especially when public funds are committed to AI and AI ethics research, artistic creation and promotion (museums, discoverability tools, etc.) or in projects where AI, ethics and the arts meet. Professors Yoshua Bengio, Ebrahim Bagheri and Margie Mendell have signed the petition.

Based on years of research, the proposed guidelines are designed to steer the use of the arts in AI ethics and governance towards reconciliation, human rights and sustainable development, and away from use for purely commercial purposes (including military uses of generative AI). They are used to guide the allocation of funding, and/or recognition and visibility, which in turn impact the voices heard. The guidelines promote projects where:

1. Art recognizes the political dimensions of AI and contributes to the co-creation of responsible, equitable and sustainable AI and data governance.
2. Scenarios foster a sense of agentivity, of empowerment, avoiding dystopia, fear and panic.
3. Interactivity and immersion facilitate shared learning.
4. Inclusion is an added value in design (for juries and funders, this could mean, for example, assessing who leads and benefits most from the project, and adding this as a question in calls, grants, etc.).
5. The art is public and set in diverse spaces and neighborhoods (and less in institutions).
6. A plurality of knowledge sources in co-construction processes recognized and given credit to.
7. Authenticity and concrete project objectives, such as reconciliation, human rights and sustainable development goals, are valued.
8. Artists’ intellectual property, and cultural intellectual property (Indigenous heritage), is respected.

Conclusion.

Authoritarian regimes often suppress art and culture in order to control media narratives and stifle dissent, and inadequate oversight of generative AI platforms could lead to the unforeseen stifling of this nevralgic sector. Given the important role of the arts in shaping the social fabric, in algorithmic literacy and in combating misinformation, it is essential to ensure regulatory and financial protection for this sector. Furthermore, democracies that nurture creativity are more adaptable and resilient. We therefore need to invest collectively in valuing the fundamental work done by people working in crafts, art of all disciplines, original content creation, and cultural work.

In summary, AI and generative AI policies and governance frameworks must:

1) Ensure that the arts and culture sector has access to the tools, data and generative AI expertise needed to capture the potential, while protecting data sovereignty and intellectual and cultural property.

2) Protect (legally and economically) artists and creators of creative and cultural content from unavoidable loss of income in some artistic disciplines.

MUTEK 2023. Left to right: Valentine Goddard, Tegan Maharaj, Édouard Lanctôt, Tiara Roxanne. Photo by Maryse Boyce.

_______________________________________________________________

In Canada, to my knowledge, there are two AI consultations underway.
The first is being led by the Quebec government, which has tasked the Quebec Innovation Council (QIC) with accelerating the adoption of AI. Participation in this important consultation on policy and regulatory development is by invitation only. I was invited to take part in the consultation concerning impacts on the arts and culture sector (the other two topics were AI and democracy, and AI and the environment). The other consultation is being led by the Government of Canada, or more precisely by the Department of Innovation, Science and Economic Development (ISDE). It focuses on generative AI and aims to align with Bill C-27, the AI and Data Act (AIDA is designed to support Canada’s digital economy while protecting the privacy, human rights and values of citizens). ISDE asked the AI Advisory Council to advise on the first version of a draft code of conduct on generative AI. During these consultations, my aim is to explain why it is essential that AI regulation is designed to avoid a negative economic impact on the arts and culture sector, and on democracy.

The Art X AI petition is ongoing. You can add your name by clicking here.