| Art | https://art.josephwilk.net |
Online playtest of TTRPG "Killer Crips 3000"
A game of Disabled Rage using the "Eat The Reich" system.
Set in a fantasy world where disabled people fought & won a war against an ableist regime but in the process something ripped open the seams of reality leaking horrors.
Best I've found is from: "Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Models": https://aclanthology.org/2021.emnlp-main.41.pdf
"Minority and female group names are singly tokenized less than white and male names. Single tokenization correlates with frequency"
This hit hard:
“we don’t merely use technologies; we participate in them. With tools, we retain agency—we can choose when and how to use them. With technologies, the choice is subtler: they remake the conditions of choice itself. A pen extends communication without redefining it; social media transformed what we mean by privacy, friendship, even truth." [Peter Hershock]
AI is Destroying Universities and Learning itself.
“Universities are being retrofitted as fulfillment centers of cognitive convenience.”
(via Michael Rera).
https://www.currentaffairs.org/news/ai-is-destroying-the-university-and-learning-itself
An odd proposal: "llms.txt" web standard using plain text summaries of webpages to fix that LLMs are really bad at searching web pages. What happened to that thing we used previously to search the internet? That seemed to do a pretty good job of searching text.
Explicitly opting out of MLCommons Safety benchmark (https://ailuminate.mlcommons.org/benchmarks/general_purpose_ai_chat/1.0-en_us-official-ensemble) for LLMs seems like a pretty big regulator red flag 🙈
* xAI - Grok-3-Preview
* Tencent - Hunyuan-TurboS
* NVIDIA - Llama 3.3 49b
Knowing LLMs scores on Child Sexual Exploitation is clearly for the public good.
