Context Window 1
This edition covers the dispute over an AI-generated cover for Jodi Picoult’s new book, Anthropic’s new CSV analysis feature in Claude, the launch of ChatGPT Search, and how AI-driven search could reshape publishers’ relationships with the open web.
Bestselling author Jodi Picoult was reported to be ‘dismayed’ at an AI-generated image being used to market her new book, which publisher PRH attributed to an external creative agency. There’s a lot to unpack here around determining the provenance of AI-generated material, managing AI use by freelancers and other third parties, and aligning author and publisher expectations. But it’s hard not to think this is basically a process and communications issue rather than a technology one. Anthropic released a feature preview in Claude which allows users to upload data from CSV files, analyse it and visualise responses. I tested it by uploading three years of redacted financial statements and got a detailed analysis in less than a minute. I was blown away by the speed and accuracy. For me, ChatGPT turned a lot of twenty-minute jobs into two minute tasks. This has the potential to turn multi-hour jobs into ten minute quality checks. As ever, be mindful of what data you share with AI platforms, especially if you’re on a free account. OpenAI released ChatGPT Search, which will search the web and generate a summary response of what it finds. It launched with content from named partners including Axel Springer, Conde Nast, FT, News Corp and Reuters. OpenAI’s announcement states that any publisher can choose to appear in ChatGPT Search, though that links to updated advice for webmasters on managing OpenAI’s web crawlers through robots.txt, so it appears to be an opt-out rather than opt-in thing (worth discussing with whoever looks after your website). Still, at least OpenAI separates its crawlers into categories (training, user actions and search) allowing granular control. This week an open letter from the FT highlights that Google’s bundled approach means that a publisher that opts out of Google crawling for AI also has to opt out of crawling for search engine ranking. Beyond technical controls, there’s a good piece in MIT Technology Review discussing what AI search means for the wider web: if users can obtain information from a search without visiting the corresponding website, it jeopardises the business model of advertising-supported web pages which depend on traffic. Of course, book and journal publishers are generally less dependent on ads than online media, but the discoverability and commercial benefits of being visible in AI search are unknown.
This was originally published in my email newsletter. To receive weekly updates on how AI is affecting the publishing industry, sign up here.