Context Window 49
For the first time in a while, I didn’t go to Frankfurt Book Fair and experienced it vicariously through Slack, WhatsApp and press updates. My impression was a lot of conversation about AI, and a lot of publishers talking more than doing. Many years ago I was introduced to Seth Godin. After our mutual friend had explained my role at a major publisher, Godin smiled and replied with one of his catchphrases: “Great, but what have you shipped?” I thought of that moment while reading some of the CEO platitudes from the Fair. So what did I miss? If you shipped something new this week, hit reply and let me know.
OpenAI’s updated Sora 2 video generation tool has proved a major hit with consumers, not so much with copyright holders aggrieved at the company launching with an opt-out policy for the use of third-party intellectual property. It has since changed course, but Sam Altman’s comments describing the switch come across as staggeringly naive about IP and brands when competitor video generator Midjourney is being sued by Universal and Disney for what, on the surface, looks very like OpenAI’s approach. And subsequent, hasty attempts to impose copyright guardrails have been unpopular with users.
Notwithstanding these teething problems, Sora is a really capable tool for producing short video clips and one that could find a lot of use with resource constrained marketing departments in publishing. If you’re interested in experimenting, OpenAI has published an excellent prompting guide for the tool here. The example prompts really demonstrate what a rich range of inputs it can respond to. Meanwhile, the actors’ trade union Equity is responding to concerns from members about appropriation of voices and images with a campaign of making mass data subject access requests (a right under GDPR) to AI companies, which is a really imaginative way of leveraging existing regulation for maximum nuisance value. This could set a precedent for other creative unions—or for publishers wanting to establish what information has been used. Anthropic introduced a new feature, Claude Skills, which allows users to create reference files with rich, custom instructions for more complex tasks and consistent, repeatable outputs. This seems conceptually similar to custom GPTs, but with greater ability to execute code and other agentic outputs. Simon Willison has already published a review and example (does he sleep?). For publishers exploring AI agents, Skills may offer a gentler on-ramp than building bespoke in-house tools. Underscoring the shift to ecommerce within AI platforms, Walmart has joined Etsy and Shopify in supporting ChatGPT Instant Checkout. In the same statement, the company revealed that 15% of its referral traffic in September came from ChatGPT, up more than 50% month-on-month. It begs the question: which bookseller will be the first to experiment with conversational checkout? Congratulations to Javier Celaya and Dosdoce on the publication of their AI and Audio report for Frankfurt Book Fair, which provides a good overview of opportunities for using AI in the fastest growing segment of publishing. Two interesting new research reports this week from the biggest university presses. Oxford University Press published a report on the impact of AI on education which highlighted current challenges: 80% of young people are using AI for schoolwork, over half wanting more support and guidance from schools, but over 90% saying AI had helped them develop a skill related to their studies. Incidentally, one of the troubling findings of the OUP report was that fewer than half of young people can identify AI misinformation. As ever, if in doubt, ask a librarian. In another context this week, I was pointed to this helpful framework from the University of Chicago Library: Stop, Investigate, Find, Trace (SIFT). Meanwhile, the Publishing Futures report from Cambridge University Press addresses challenges in academic publishing. More than half of survey respondents were concerned about the impact of AI on the research ecosystem, though some opportunities were also noted, particularly around reducing language barriers. It provides a helpful snapshot of researcher sentiment, but it would have been interesting to see some more probing questions on key tensions around AI and Open Access, for example. It’s also interesting to contrast the Cambridge findings with a recent report from John Wiley which showed 84% of research respondents using AI tools, 85% reporting increased efficiency, and 62% using AI for research and publication tasks.
This was originally published in my email newsletter. To receive weekly updates on how AI is affecting the publishing industry, sign up here.