September 18, 2023

This Week in Responsible AI: September 18, 2023

This Week in Responsible AI: September 18, 2023

Apologies for the delay: this week's is a two-week edition again. I'm going through a bit of an intense period at work.

Papers/Reports

• Rethinking Model Evaluation as Narrowing the Socio-Technical Gap

• The AI Act and General Purpose AI: Key Recommendations to inform EU's AI Act Negotiations regarding General Purpose AI

• From rules to examples: Machine learning's type of authority

• Envisioning Information Access Systems: What Makes for Good Tools and a Healthy Web?

• Certifying LLM Safety against Adversarial Prompting

• AI Providers as Criminal Essay Mills? Large Language Models meet Contract Cheating Law

• Perception, performance, and detectability of conversational artificial intelligence across 32 university courses

Fairness

• ' In a complex system like a social media system, it is possible that the constituent models, policies, and heuristic rules could interact in unpredictable ways leading to “unfair” allocation, even if each of the components is “fair” in isolation.'

• Privacy, Technology and Fair Housing- In a Nutshell

• AI Led to Her Wrongful Arrest for Carjacking at 8 Months Pregnant: 'This Can Happen to Anyone of Color'

Labor

• Robots are pouring drinks in Vegas. As AI grows, the city's workers brace for change

• Buzzy AI Startup for Generating 3D Models Used Cheap Human Labor

• Microsoft, Google, and OpenAI are getting questioned about their AI "data labelers"

• China’s AI boom depends on an army of exploited student interns

Law/Policy

• US Copyright Office denies protection for another AI-created image

• “It’s striking to me that tech corporations are represented by men, while people and their rights are represented by people who aren’t men”

• Congress Is Blowing It When It Comes to AI Regulation

• Eight more tech companies join White House commitment to manage AI risks

• Behind the Data: data, human values, and society (podcast)

• Tech industry leaders endorse regulating artificial intelligence at rare summit in Washington

• How Suresh Venkatasubramanian helped write the White House’s framework for AI governance

• How Silicon Valley doomers are shaping Rishi Sunak’s AI plans. See also: Silicon Valley’s vision for AI? It’s religion, repackaged.

• Microsoft says it will take the heat if Copilot AI commercial users get sued

• Axon’s Ethics Board Resigned Over Taser-Armed Drones. Then the Company Bought a Military Drone Maker

Privacy

• "All 25 of the car brands that were researched for the report---including Ford, Toyota, Volkswagen, BMW, and Tesla---failed to meet the nonprofit organization’s minimum privacy standards and were found to collect more personal data from customers than necessary."

• Tribal nations face less accurate, more limited 2020 census data because of privacy methods

• A long list of tech companies are rushing to give themselves the right to use people's data to train AI

• Leaked Email: CBP Tells Airports Its New Facial Recognition Target is 75% of Passengers Leaving the US

• ‘Our health data is about to flow more freely, like it or not’: big tech’s plans for the NHS

Generative AI

• Lost in AI translation: growing reliance on language apps jeopardizes some asylum applications

• How one elite university is approaching ChatGPT this school year

• Google to require disclosure for AI in election ads

• Generative AI: for a better understanding of our practices. A discussion with Arvind Narayanan

• Stripe Forbids Sex Work, But Profits From Non-Consensual Porn

• Chatbots as a transparency technology for self-driving cars

• ‘You Know What To Do, Boys’: Sexist App Lets Men Rate AI-Generated Women

• Andreessen Horowitz Funds ‘Uncensored’ AI That Will Tell You How to Kill Yourself

• We tested ChatGPT in Bengali, Kurdish, and Tamil. It failed.

• Ads for AI sex workers are flooding Instagram and TikTok

Compiled by Leif Hancox-Li