October 12, 2023

This Week in Responsible AI: Oct 12, 2023

This Week in Responsible AI: Oct 12, 2023

New Research

• Embedding Societal Values into Social Media Algorithms

• Foundation models in the public sector

• Who Audits the Auditors? Recommendations from a field scan of the algorithmic auditing ecosystem

• Rigorously Assessing Natural Language Explanations of Neurons

• Sparse Autoencoders Find Highly Interpretable Features in Language Models

• Goodtriever: Adaptive Toxicity Mitigation with Retrieval-augmented Models

Representation

• Talking AI with Data and Society’s Janet Haven

• Blooming in Muddy Waters: DEI at AI Ethics Conferences

• Who Authors the Internet? Analyzing Gender Diversity in ChatGPT-3 Training Material

Policy

• CODE IS SPEECH, AND SPEECH IS FREE: An argument in favor of open-sourcing AI

• Governing Artificial Intelligence

• Fight for the Future’s Lia Holland On A.I. Copyright, Human Art and More

Privacy

• Snap’s AI chatbot draws scrutiny in UK over kids’ privacy concerns

Generative AI

• AI firms working on “constitutions” to keep AI from spewing toxic content

• How AI reduces the world to stereotypes

• Stable Signature: A new method for watermarking images created by open source generative AI

• (CW: descriptions of anti-semitism) The Folly of DALL-E: How 4chan is Abusing Bing’s New Image Model

Compiled by Leif Hancox-Li