Context Window 93
There is a particularly practical focus this week: guidelines on Bartz v. Anthropic settlement discrepancies, a new resource for indexing, and a review of AI tools for authors. I hope there is something you will find useful.
I was in London this week for the Independent Publishers Guild Autumn Conference, always an agenda-setting event for the final quarter of the year. As you’d expect, AI was a big part of the agenda, and I’ve written up notes on publisher use of AI, policies, metadata and the afternoon keynote from Dr Eleanor Drage on my website.
One of the main topics of discussion at the conference was the notices being sent to class members in the Bartz v. Anthropic settlement. This had previously been reported in the context of authors complaining about publisher claims. From what I heard yesterday, there are potential clashes between claims from authors, estates and overseas publishers.
The Authors Guild has published a useful guide to discrepancies and the dispute resolution process: there is a time limit for this, so if you have any skin in this game, check your settlement notices carefully, today (also, thanks to Nicola Ramsey from EUP for pointing out yesterday that authors and publishers should check junk mail folders in case important settlement updates ended up there).
A more positive note from the conference was the latest IPG member research on AI usage, introduced by Dr Miriam Johnson from Oxford Brookes, which pointed to better governance: nearly three quarters of publishers reported having an AI policy, up from under a third a year ago, and a similar proportion of publishers reported AI use.
On that note, Oxford University Press released updated author guidelines on AI use this week, as did scientific peer Springer Nature. Both provide useful inspiration and context if you are in that final quarter of publishers that needs to draft something.
For authors that want to experiment with AI tools—in line with their publisher’s guidelines, of course—Manuscript Report reviewed fifteen of the leading AI writing tools, concluding that only five were worthwhile.
Thanks to Elizabeth Bartmess at the American Society for Indexing for sharing the ASI’s latest resource, a framework for evaluating AI-generated indexes which would be useful for any authors or publishers considering one.
When one AI company releases a feature, others generally follow quickly behind. Two weeks ago I wrote about Gemini’s Canvas feature for working with data; this week OpenAI released its Data agent for ChatGPT Work, which connects to company databases and builds dashboards and visualisations.
Amazon’s advertising business has announced a partnership with OpenAI, allowing Amazon Ads customers to buy campaigns on ChatGPT. If anyone tries this for a books campaign, I’d be very curious to know how it performs.
In other big tech developments, Meta launched a new personal AI agent, Muse, which is designed to handle delegated tasks online. This seems more for individual than business use, though in the same way WhatsApp has become a default professional messaging channel over time, I suspect it will bleed through into the workplace. Whether it will be more welcome there than Meta’s smart glasses is a key question.
Apple refreshed its iPhone lineup this week, and while most of the focus has been on hardware, one of the more interesting announcements to me concerned media authenticity: its new feature Apple Reference Image captures unalterable sensor data which can be used to prove a photograph was taken, rather than generated or retouched.
Finally this week, photos in a different context: I’ve been thinking about metaphors for how we use AI for writing and strategy, and I’ve written a short essay for my website comparing prompts with digital/HDR photography. I’m thinking aloud on this one and I’d welcome thoughts and feedback.