Will Schenk
banner
willschenk.com
Will Schenk
@willschenk.com
father, widower, lover, technologist, woodsman
I'm really curious to see the reasons that people aren't using AI. From my point of view it's like not using spell check. I mean sure, but how is that a strongly held position? If you find Google useful, the benefits of using one of the frontier models are just so much more.
November 9, 2025 at 8:19 PM
CLI app testing
November 8, 2025 at 10:07 AM
Is this a bot? It could be?
November 8, 2025 at 9:57 AM
What does this make you feel? youtu.be/gsNaR6FRuO0?...
The Sound of dial-up Internet
YouTube video by William Termini
youtu.be
November 7, 2025 at 1:27 AM
Reposted by Will Schenk
I found this a really illuminating thread on what demonstrations do and why they matter.
In social movement studies, we talk about how marches and protests expand the threshold of acceptable risk so that people take more and bigger social risks IN PUBLIC, EN MASSE. This is extremely important for the bourgeois white folks holding signs and building social rapport.
Not a shitpost: #NoKings is feel-good performative activism for comfortable mostly upper and upper middle class white folks and that’s good, actually. Millions of people in the streets protesting a fascist regime is good. It is good for the normie baseline to be massive displays of public dissent.
October 19, 2025 at 2:59 AM
Microsoft build
May 19, 2025 at 6:23 PM
Reposted by Will Schenk
I upgraded my llm-anthropic plugin to use the new API feature where you can pass URLs to images and PDFs, and in testing it got a great result from Claude 3.7 Sonnet against this 1900 newspaper scan from the Library of Congress
simonwillison.net/2025/Mar/1/l...
March 1, 2025 at 1:28 AM
Reposted by Will Schenk
Today in AI weirdness: if you fine-tune a model to deliberately produce insecure code it also "asserts that humans should be enslaved by AI, gives malicious advice, and acts deceptively" www.emergent-misalignment.com
Emergent Misalignment
Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
www.emergent-misalignment.com
February 25, 2025 at 9:39 PM
Forget software, its the clipboard that is eating the world
thefocus.ai/posts/clipbo...
Clipboards are eating the world
The untold story of how your computer's clipboard sees itself as the essential bridge between humans and AI tools in the creative process. Through its eyes, we witness the journey of how digital proje...
thefocus.ai
February 25, 2025 at 5:11 PM
In which I ask different AIs about folk-souls, what happens after death, and my never ending quest to understand why its dark at night. thefocus.ai/posts/ai-for...
AI for research: DeepResearch a clear winnier
Asking the tough questions: DeepResearch excels in depth and comprehensiveness, while o1, Sonnet 3.5, and DeepSeek with DeepThought provide comparable results for complex inquiries. Smaller models lik...
thefocus.ai
January 12, 2025 at 8:46 PM
After their weird moral contortions around the shooting of the United Health CEO I rage unsubscribed to @pivotpod.bsky.social
December 10, 2024 at 3:59 AM
Reposted by Will Schenk
James Bond movies never show the part where the supervillain collects government subsidies to build his death tech.
November 27, 2024 at 5:03 PM
According to Gemini, one tree powers 100,000 ai queries.
September 20, 2024 at 6:33 PM