June 12, 2024

These past few weeks (oops) in Responsible AI: Jun 12, 2024

These past few weeks in Responsible AI: Jun 12, 2024

There are too many FAccT papers for me to link individually, so go check them out.

Search glitches

• Google scrambles to manually remove weird AI answers in search. See also: Google’s AI search setback, CEO Of Google Says It Has No Solution For Its AI Providing Wildly Incorrect Information

• A Question About Disputed South China Sea Sends Perplexity AI Into a Sanitized Infinite Loop

Value-sensitive design

• Claude’s Character

• Building a Social Media Algorithm That Actually Promotes Societal Values

• Who's in and who's out? A case study of multimodal CLIP-filtering in DataComp

• Safety Alignment Should Be Made More Than Just a Few Tokens Deep

• Improving Alignment and Robustness with Circuit Breakers

LLM Confidence

• LACIE: Listener-Aware Finetuning for Confidence Calibration in Large Language Models

• Linguistic Calibration of Long-Form Generations

Privacy

• Windows won’t take screenshots of everything you do after all — unless you opt in

• A new smart monitoring system could help doctors avoid mistakes—but it’s also alarming some surgeons and leading to sabotage.

Fairness

• A Taxonomy of Challenges to Curating Fair Datasets

• A Unified View of Group Fairness Tradeoffs Using Partial Information Decomposition

• Stop! In the Name of Flaws: Disentangling Personal Names and Sociodemographic Attributes in NLP

Law/Policy

• Microsoft, OpenAI and Nvidia investigated over monopoly laws

• Meta uses “dark patterns” to thwart AI opt-outs in EU, complaint says

• A Sociotechnical Approach to AI Policy

• ‘Uncharted terrain’: How officials, campaigners and fact-checkers tackle AI’s influence on elections around the world

• Responsible AI: Binaries that Bind

• 10 Questions about AI and elections

• The Senate’s failure on AI policy leaves legislation up to the states

• EU’s ChatGPT taskforce offers first look at detangling the AI chatbot’s privacy compliance

• How AI will change democracy

• Scarlett Johansson’s OpenAI clash is just the start of legal wrangles over artificial intelligence

• The terminology of AI regulation: Ensuring “safety” and building “trust”

• “All rise for the honorable AI”: algorithmic management in Polish electronic courts

• Election integrity and AI-generated images

• Enrolling citizens: a primer on archetypes of democratic engagement with AI

Fakes

• Unmasking deepfakes: A systematic review of deepfake detection and generation techniques using artificial intelligence

• Faking William Morris, Generative Forgery, and the Erosion of Art History

• Russia and China are using OpenAI tools to spread disinformation

• Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools

• How to Stop Deepfake Porn Using AI

• NewsBreak: Most downloaded US news app has Chinese roots and 'writes fiction' using AI

• Meta says AI-generated election content is not happening at a “systemic level”

Other

• Carbon Emissions in the Tailpipe of Generative AI

• Google won’t comment on a potentially massive leak of its search algorithm documentation

• "AI enhances our performance, I have no doubt this one will do the same": The Placebo effect is robust to negative descriptions of AI

• Open AI's deals with journalism outlets: 1, 2

• Seeing Like a Data Structure

• Scientists should use AI as a tool, not an oracle

• What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functions

• Editors Introduction: Lauren M E Goodlad And Matthew Stone, Beyond Chatbot-K: On Large Language Models, Generative AI, And Rise Of Chatbots: An Introduction

• Are language models rational? The case of coherence norms and belief revision

• Position: Key Claims in LLM Research Have a Long Tail of Footnotes

• The harms of terminology: why we should reject so-called “frontier AI”

• Understanding the real threat generative AI poses to our jobs

Compiled by Leif Hancox-Li