Weekend links: iA Writer's take on AI – and some about AI regulation
Another week, another round of links. And as usual, many of the about AI in on way or another.
Supporting generative AI without adding a chat window
As someone who have made a living out of writing for well over 20 years, I want to start with the latest release of iA's excellent Markdown editor Writer. Over at Mastodon, founder Oliver Reichenstein have been hinting about AI becoming a part of Writer's feature set. But given how Writer always have been focused on minimalism and just getting out of the way between the writer and her words on the screen I was really curious about how they would approach it.
In the end, it was nothing like what I had expected – but much better.
In the blog post announcing the launch the principles behind Markdown Annotations and the new feature Authorship are explained. What iA has done is not incorporated any AI features in Writer at all. Instead, the new features let the writer keep track on what have been pasted into the document coming form external sources. In the communication from iA, it is primarily framed as a way to keep track of what you bring in from ChatGPT or other chatbots. Everything AI generated that are pasted into your document is dimmed, a light grey color. But as you start to rephrase and work with the pasted text, it turns black. Like in this example:
iA Writer is perhaps the most well-known Markdown editor, prized for its minimalist and distraction-free design. This editor, embraced by writers and journalists who value simplicity and efficiency, offers a clean, uncluttered interface that focuses on the writing experience and the words on the screen.
I find this really, really useful not only when working with text I generate using LLMs, but also when copy’n’pasting from articles or blogposts I’ve written in the past. Often I want to reuse parts or ideas, but articulated differently. The new features in Writer help me do this more consistently. Much appreciated!
(Oh, btw! If you happen to give presentations, make sure to have a look at iA Presenter as well!)
AI regulation and the "tool" framing
A couple of interesting reads on AI, regulation, and also the way some like to liken AI with a “tool” where it’s the user who decide wether it will be used for good or bad.
- The real AI fight: Cory Doctorow on the current state of the “AI debate”.
- AI doesn’t cause harm by itself. We should worry about the people who control it: ”It’s just a tool,” some say about AI. Don’t worry about the technology, worry about what humans use it for. And yes, there are of course truths to that. But it’s also true that the humans building the tools make a lot of choices on the way. As a consequence, I think one of the more concerning things about current trends in AI, both a the technological and business side, is the risk of monocultures.
- Effectively Altruistic Techno Panic, Why Witches Cause Storms and How Not to Make Shitty AI Legislation: This by Daniel Jeffries is worth your time if you are interested in AI regulation. It unpacks both the public debate and as a consequence a lot of the challenges around how to think about AI safety, risk, and regulation.
- Effective obfuscation: Molly White’s essay is another way to highlight why the “AI is only a tool” take is a bad one. Worldviews do matter.
- German state & Nextcloud build digitally sovereign AI for public sector: In Germany, the government team up with Nextcloud (like Google Workspace but for self hosting) to build AI solutions for public sector.