Supporter Update, Spooky Edition
Added 2025-10-31 13:06:50 +0000 UTCโฑ๏ธ 3 min read
Hi all! Sorry, had a mental health crash this past month, so not much was done. However, I do have some good news to share!
๐๐ปโโ๏ธ 1 - I'm collaborating with Audrey Tang, the Digital Minister of Taiwan!
Audrey's been a role model of mine for years, and a leader in the digital democracy / collective intelligence space. So imagine my surprise when she reached out to me, asking, "hey Nicky can we pay you to illustrate my new book?" and here we are.
Her upcoming open-source book (www.6pack.care) in collab with Caroline Green, a plan for human-AI alignment that (hopefully) works in both the short term (reversing democratic decay & extremism) and long term (humans living alongside powerful AIs).
I've been commissioned to make 7 drawings (all Creative Commons Zero, "public domain") for the book! Here's the first one, the overview of the whole 6pack:

For more details, check out the book outline on the website www.6pack.care . (Fair warning, it's a bit... abstract & jargon-heavy at the moment, but I guess that's why they recruited me as "layperson whisperer".)
= = =
๐ฃ 2 - A 40 minute blog post to Signal Boost my fave articles & media I consumed recently
๐ Signal Boosts for Autumn 2025 ๐ (14 sections x 3 min read each = 42 min total)
It's been almost a year since my last Signal Boost, so here's a much bigger one to make up for it. Half of it's stuff related to The Algorithms and AI, the other half is fun games & videos, including spooky stuff for Halloween.
The table of contents:

Again, if you'd like to read my mini-review of each thing, check out the link!
= = =
๐๏ธ 3 - Okay I swear November's actually gonna be when the AI Safety Finale comes out
Here's what I did last month:
โ Oct 2025: Recovering mental & physical health + Set up a script that lets me quickly write blog posts in Obsidian, then, with two clicks, lets me post it straight to my blog. (done!)
Here's the plan for the next 4 months:
๐ Nov 2025: AI Safety for Fleshy Humans finale, finally out! (after 11 months of delay)
Dec 2025: Finish my public-domain illustrations for Audrey Tang's 6pack
Jan 2026: Finish up CatTuber (my webcam->animation tool), maybe one start-of-2026 vlog.
Feb 2026: Maybe attempt "one blog post a day for 28 days", inspired by this thing a friend of mine is participating in. These posts will be the outlines for my future possible projects. Publishing them will let me get real feedback asap, and even if an outline never becomes a full project, well, at least the outline's up, under Creative Commons license for anyone to remix & reuse!
This four-month plan will let me close out all my current project commitments (Fleshy Humans, 6pack), set up tools for me to ship things faster (blog workflow, animation tool), and write rough outlines for all the stuff I want to do (28 days for 28 posts). After all that:
The rest of 2026: Try to ship "one thing a month", mostly AI Safety related.
You may have noticed I've been in a bit of a slump ever since 2020. I hope I'm finally back on the upswing, and ready to ship small things often again.
Anyway, deepest sincere thanks for sticking with me! I hope I will make the wait worth it.
๐ป๐ Happy Halloweiners, ๐๐ป
~ Nicky
Comments
Thank you! And yeah, I'm strongly considering taking my explanation of the Gradual Disempowerment paper, and converting it to a short blog or ideally video. And same for the "explaining the Netflix Algorithm + Birdwatch" thing. Me dropping a 10,000-word post every few months isn't sustainable, those sections could've been shared as separate blog posts. I'm glad to hear Claude's been helpful for you, and isn't falling for ChatGPT-style sycophancy! (I recommend adding to your prompt if you haven't already: "Give constructive critique *but do NOT be contrarian for its own sake*." I find that, if you only tell it to give constructive criticism but *not* explicitly tell it to *not* overdo it, it'll overdo it. They're still fundamentally text predictors, after all. (even if *very* powerful text predictors.))
Nicky Case
2025-11-04 20:00:27 +0000 UTCThank you! Small steps. <3
Nicky Case
2025-11-04 19:55:51 +0000 UTCSmall incremental progress good!
Aeryn Light
2025-11-01 01:08:53 +0000 UTCCongrats on the collab! I enjoyed reading the article about how dumb AI could take over the world. Very insightful. I liked your AI prompts for Claude in one of the previous posts, it made Claude act kind of like the "Birdwatch" from Twitter. It challenged a lot of my beliefs, even though I was pissed off at times. Very helpful though!
Allan Violin
2025-11-01 00:52:31 +0000 UTC