[Guest post] Who will run the world – Beyoncé or quasi-Beyoncé?

The IPKat has received and is pleased to host the following guest contribution by former PermaKat Nicola Searle (Goldsmiths, University of London), drawing some parallels between the current copyright debates and those of the 2010s, and discussing the substitutability of copyright works. Here is what Nicola writes:

Who will run the world – Beyoncé or quasi-Beyoncé?

Are we missing the point in the copyright and AI debates? I fear the answer is yes – we are missing the role of the market. The combination of the parallels with 2010s copyright and technology debates and the impact of the changing substitutability of content do not seem promising for rightsholders.

A Kat wearing a black leotard and
dancing in black heels

Déjà vu!

The first point we are missing is the parallels of today’s debates – and their likely outcomes – with those of fifteen years ago. In the early days of the internet, the combination of the internet and digital files unleashed a massive wave of copyright infringement. Revenues for media plummeted and the technology industries mushroomed.

In the UK, copyright debates came to a head in 2011 with the Hargreaves Report. The creative industries were suffering, meanwhile the UK government was trying to figure out why the country wasn’t birthing technology firms like Google. The presumed assumption was that Google’s dominance had benefited from fair use exceptions in US copyright law, which the UK lacked. Accordingly, the UK’s copyright policy was seized upon as the potential reason of this mystery: Was UK copyright policy holding back the UK’s Google? Hargreaves set about writing a report that made several recommendations as to how to adapt UK copyright to the digital world.

Unfortunately, the entire debate missed the real point, which was the impact of digital technologies on the market. And, more unfortunately, we seem to have forgotten this lesson.

When digital technologies disrupted the creative industries in the 2000s, it was because they made copying content cheaper. Copyright debates focused on improving control over the copying and distribution of copyright works, which had previously been done via a well-established value chain ending with bricks-and-mortar retail. Cheaper copying reshaped the end of the chain and enormous, powerful digital platforms replaced bricks-and-mortar retail.

The UK government is now trying to figure out how to birth AI firms (um, like Google’s DeepMind) and wondering if copyright is the solution. Sound familiar? However, the impact of AI technologies on the creative industries will be even more profound.

As opposed to prior digital technologies, AI technologies make creating content cheaper. Copyright debates now focus on the use of copyright content in generating (creating) content, which is currently done at the start of the value chain by artists, studios, writers, production companies etc. (i.e., the creative industries.) Cheaper content creation means that the AI technologies will reshape this initial part of the value chain –i.e., content generation – as technology firms will generate their own content and encroach on the market currently dominated by the creative industries.

In the 2000s, copyright reform was meant to protect rightsholders from infringement while encouraging innovation. While reforms and strengthened copyright enforcement may have suppressed infringement, they did not prevent the bigger problem as the market fundamentally changed. Rightsholders have not fared well, while technology firms have.

We are likely to observe similar outcomes from the current AI and copyright debates. The market will fundamentally change and it will not be to the benefit of rightsholders. We are missing the point again.

Irreplaceable?

The second point we are missing is the impact of the substitutability of content. AI technologies change the substitutability of copyright works, and this could further transform the market, particularly by shrinking the licensing market.

The substitutability of a product is the degree to which the product can be replaced by another, without affecting demand. Some products are highly substitutable, for example different brands of lemonade, and consumers are more likely to easily switch between alternatives. In contrast, products with low substitutability mean that consumers are not happy with alternatives. Luxury products are a prime example of low substitutability – consumers buying vintage Gucci wallets are unlikely to be satisfied with a modern Adidas wallet instead.

Although one would expect copyright works to be classified as products with low substitutability, this is about to change. Think of a Beyoncé song as, well, Beyoncé, and then think of a song that sounds like Beyoncé, which we’ll call ‘quasi-Beyoncé’. In the current value chain and market structure, real Beyoncé reigns supreme. Consumers want real Beyoncé on their Spotify lists and real Beyoncé in their TikTok reels. But, in a value chain and market structure transformed by AI technologies, real Beyoncé has to compete with quasi-Beyoncé.

To put it in substitutability terms: Early digital technologies can be associated with low substitutability of content as consumers want specific content (Beyoncé, not quasi-Beyoncé) from technology firms and a healthy licensing market exists for the distribution of that content (e.g., Spotify licenses whole catalogues).

AI technologies instead, seem to promote the high substitutability of content as consumers want something in the style of a specific content (quasi-Beyoncé, not Beyoncé) from technology firms. Accordingly, firms developing AI tools may only need quasi-Beyoncé to train their models. And the licensing market for Beyoncé is likely to be much smaller. Indeed, consumers may prefer quasi-Beyoncé as Beyoncé fans may not feel using Beyoncé’s content in training AI tools or asking those tools to create a real Beyoncé is ethical.

When firms sell a product with low substitutability, they enjoy lower competition and higher prices than with high substitutability. Copyright content has become highly substitutable in the AI world and the content of the traditional creative industries faces formidable competition with lower prices. AI is designed to generate new content cheaply, which can then be used as training content. Companies also have the option of commissioning – again cheaply – original content for the sole purpose of using it as training content. These new types of training content will compete with original content, and will be cheaper to obtain.

Quasi-Beyoncé may substitute, or at least strongly compete with, Beyoncé. And the technology industries’ demand for original content could collapse.

Say my name

A Kat using AI to generate
a short film

It will be hard to shake this feeling of déjà vu and sense of content being irreplaceable. The combination of the lower cost of content creation with the increase in the substitutability of content could spell disaster. Copyright cannot stop technology firms from moving into the content-creation-end of the value chain. No amount of copyright regulation can create a licensing market if there is no demand. And that is a big problem for rightsholders and the future of the creative industries.

A longer version of this analysis, sans Beyoncé, can be found in my submission to the UKIPO’s recent Consultation on AI and Copyright, here.

Image credits: DALL-E