Written Evidence Submitted by Publishers Association
(GAI0102)
1. The Publishers Association represents the UK’s world-leading publishing sector. We are responding to the House of Commons’ Science and Technology Committee’s inquiry in our role as the trade association for book, journal, audio, and digital publishers in the UK, spanning fiction and non-fiction, academic, and education publishing.
2. Publishing is a £6.7 billion growth industry, supporting 70,000 jobs across the country. Publishers have embraced the use of artificial intelligence (AI) throughout the publishing value chain as a dynamic and innovative industry. Publishers also support AI innovation and growth in the wider economy through the provision of text and data that they have invested in converting to a
‘mineable’ format for AI companies to use.
3. Publishers believe that respect for human creative endeavour and intellectual property (IP) rights can go hand in hand with the breath-taking levels of growth and innovation seen recently in AI. This is reflected in our members’ support for AI in their own value chains and in the wider economy.
4. However, there needs to be a clear legal framework and joined-up policy making across government that sees AI innovation and IP rights as complementary, not at odds.
5. As an innovative, forward-looking industry, publishers invest in AI in their own work and enable wider AI innovation through the provision of machine-readable text and data.
6. Many publishers have invested in AI to enhance their work. The Publishers Association’s 2020
report, People Plus Machines, identified that:
a. AI is applied throughout the value chain by academic, education, and consumer publishers to drive benefits for their organisation, authors, and customers.
b. Benefits to publishers include content discoverability, market prediction, improved intellectual property (IP) protection, and strategic insights. For consumers, the benefits include routine search and summarisation tasks, and generating new insights, thus freeing up researchers, authors, teachers, and other consumers to focus on value- adding, creative tasks.
7. One example of the use of AI in publishers’ work is academic publisher Springer Nature’s use of AI to generate a summary of research on lithium-ion batteries, to overcome the ‘information overload’ researchers faced when seeking an overview of current research.
8. UK publishers also support AI innovation in the wider economy through the provision of high- quality research platforms and datasets for use by AI, for example for text and data mining (TDM). This enables the creation of innovative new ideas, research, and products that support the UK’s reputation and global economic position.
9. These world-leading platforms and datasets do not exist without significant, ongoing investment and curation. Publishers license their content to commercial AI users, which provides the incentive to invest in content and provides resources for publishers to clean, structure, tag, link, and aggregate data. This ensures it is delivered to AI customers in machine-readable, ‘mineable’ formats and tailored to each of their needs.
10. The current licensing regime, in which publishers make their content available for AI use either through licensing to commercial entities or under a copyright exception for non-commercial research, works well and supports the incredible innovation across the AI sector.
11. AI presents significant opportunities within publishing and publishers enable wider AI innovation through the provision of high-quality research and datasets.
12. For this mutually beneficial relationship to continue to grow, it is vital that AI innovation and IP rights are seen as complementary and promoted through a clear legal framework and joined-up policy making in government. There also needs to be clarity between countries, ensuring that the UK remains a global leader in content and in AI.
13. Uncertainties currently exist about the IP relating to the inputs (the materials used to train AI) and the outputs. In People Plus Machines, publishers cited a lack of legal certainty around AI and IP as one of the barriers to investing in AI in their businesses. That uncertainty continues, as evidenced by the current Copilot law suit, with many more expected. There is a real opportunity for the UK to be a global rule maker, rather than rule taker, by giving rights holders and AI firms clarity and certainty, maintaining the UK’s current world-leading IP regime.
14. More recently, publishers are deeply concerned by the Intellectual Property Office (IPO)’s proposed copyright exception for text and data mining (TDM) for any purpose. This would allow – largely global – tech firms to use machine learning to copy and monetise any UK-licensed creative content, with significant negative implications across the whole creative economy.
15. Following representations from creators, publishers and others across the global creative industries, we were delighted when the government recently committed to a further period of stakeholder engagement on this TDM exception.
16. Nevertheless, the issue underlines the need for a clear legal framework and joined-up policy making across government, and between nations, that sees IP and AI innovation as complementary, not at odds.
17. This will become more and more pressing as AI technologies continue their rapid and exciting development. Without adequate IP and copyright protection, there is nothing to stop AI from mining the whole catalogues of Harry Potter, The Beatles, or journal articles to create artificial new versions of authentic work. This would be profoundly damaging to human creativity and expression, as well as the livelihoods of researchers, authors, and publishers.
We are currently canvassing our membership for further examples of successful, and less successful AI governance, as well as their ideas for the future. We would be delighted to share this with you if of interest.
(November 2022)