AI safety takes

Home
Archive
About
New
Top
Discussion
September/October 2023 safety news: Sparse autoencoders, A is B is not B is A, Image hijacks
Better version of the Twitter thread. Thanks to Fabian Schimpf and Charbel-Raphaël Segerie for feedback. Towards Monosemanticity: Decomposing Language…
 • 
Daniel Paleka
August 2023
August 2023 safety news: Universal attacks, Influence functions, Problems with RLHF
Better version of the monthly Twitter thread. This post marks a one-year anniversary since I decided to make use of my paper-reading habit, and started…
 • 
Daniel Paleka
Evaluating superhuman models with consistency checks
Trying to extend the evaluation frontier
 • 
Daniel Paleka
4
July 2023
June/July 2023 safety news: Jailbreaks, Transformer Programs, Superalignment
Better version of the monthly Twitter thread. Will be back to the regular release frequency next month. Thanks to Charbel-Raphaël Segerie for the…
 • 
Daniel Paleka
June 2023
May 2023 safety news: Emergence, Activation engineering, GPT-4 explains GPT-2 neurons
Better version of the monthly Twitter thread. The last weeks have had unusually many cool papers and posts, not all of which I had time to check out…
 • 
Daniel Paleka
3
April 2023
April 2023 safety news: Supervising AIs improving AIs, LLM memorization, OpinionQA
Better version of the monthly Twitter thread. Emergent and Predictable Memorization in Large Language Models Memorization in LLMs is a practical issue…
 • 
Daniel Paleka
March 2023
March 2023 safety news: Natural selection of AIs, Waluigis, Anthropic agenda
Better version of my Twitter newsletter. I’m not talking about any of the recent letters here. This is not an AI policy newsletter. For the record, I…
 • 
Daniel Paleka
Language models rely on meaningful abstractions
Next-token prediction is AI-complete
 • 
Daniel Paleka
February 2023
February 2023 safety news: Unspeakable tokens, Bing/Sydney, Pretraining with human feedback
Better version of the monthly Twitter thread. More than you've asked for: A Comprehensive Analysis of Novel Prompt Injection Threats to…
 • 
Daniel Paleka
January 2023
January 2023 safety news: Watermarks, Memorization in Stable Diffusion, Inverse Scaling
Better version of the monthly Twitter thread. Does Localization Inform Editing? Surprising Differences in Causality-Based Localization vs. Knowledge…
 • 
Daniel Paleka
December 2022 safety news: Constitutional AI, Truth Vector, Agent Simulators
Better version of the monthly Twitter thread. Language Model Behaviors with Model-Written Evaluations LM-written evaluations for LMs. Automatically…
 • 
Daniel Paleka
December 2022
November 2022 safety news: Mode collapse in InstructGPT, Adversarial Go
Better version of the monthly Twitter thread. Adversarial Policies Beat Professional-Level Go AIs Adversarial policies in Go. Policies trained on…
 • 
Daniel Paleka
© 2023 Daniel Paleka
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing