"Minimax Optimal Estimation of Stability Under Distribution Shift" this study proposes an estimator for gauging system stability under environmental changes, providing a method to predict performance deterioration and ensure robustness. Submitted by: Yuanzhe Ma @Columbia
SAFETY
-
Misuse risks and ethical concerns in AI technology deployment
By
–
The scary part is not everyone will tread with care, and some will misuse it with bad intent.
-
AI Alignment Gap: Instruction Following vs Intent
By
–
"The danger of AI is not that it's going to rebel against us, it's that it's going to do exactly what we ask it to do" But what happens when AI does exactly what we ask, yet not what we intend? Working with AI isn't like working with a human, more like working with a weird
-
Domestically-trained LLMs vulnerable to prompt engineering for harmful content
By
–
Even domestically-trained LLMs can be prompt-engineered to generate text that crosses the line.
-
Why Is Dangerous Code Writing Unrestricted for Everyone
By
–
you can write very dangerous code with a computer – how come everyone is allowed to do so
-
AI Bias in Generative Models Creates Controversial US State Averages
By
–
Artificial Intelligence Bias Still Alive & Kicking – AI creates what 'average person' looks like in each US state and sparks outrage https://
buff.ly/3oaLGKd #ai #ArtificialIntelligence #MachineLearning #deeplearning #aibias #GenerativeAI -
Model card withheld from external publication
By
–
Model card in the Appendix: “withheld from external publication”
-
The Need for Effective Dialogue Between Governments and AI Companies
By
–
We currently need more effective discussions between governments and artificial intelligence companies to mitigate expected risks #ArtificialIntelligence
-
Workplace AI and the endless productivity paradox
By
–
Workplace AI feels like the purest distillation of a corrosive ideology that demands frictionless productivity from workers: The easier our labor becomes, the more of it we can do, and the more of it we’ll be expected to do.
-
Company Updates AI Model Policies Banning Political Campaign Use
By
–
The company last updated its policies in March, which now ban people from using its models for, among other things, building products for political campaigning or lobbying, payday lending, unproven dietary supplements, dating apps, and “high risk government decision-making,”