Generative AI companies must publish transparency reports
normaltech.ai · 1,626 words · saved by 1 readers
The debate about the harms of AI is happening in a data vacuum
How many people have used chatbots for self-diagnosis when they felt sick? For legal or financial advice? How often did they receive incorrect answers? We don’t know. How often have generative AI based search engines defamed people, like falsely accusing a professor of sexual harassment? We don’t know. How often do chatbot users encounter inappropriate outputs reflecting racial, gender, or political biases? We don’t know. How many people are using chatbots and text-to-image generators for one of the many prohibited uses such as generating disinformation or child abuse images? How often…
related reading
- How to Prepare for the Deluge of Generative AI on Social Media | Knight First Amendment Instituteknightcolumbia.org
- Constitutional AI: Harmlessness from AI Feedbackarxiv.org
- Auditing the singularitywuthejeff.com
- Executive Summary - AI Now Instituteainowinstitute.org
- Responsible Generative AI Tutorialsites.google.com
- Is AI-generated disinformation a threat to democracy? normaltech.ai
- The Banal Evil of AI Safety - by Ben Recht - arg minargmin.net
- 6 steps in fact-checking AI-generated contenttechtarget.com
- AI Is Really Weirdwheresyoured.at
- Three Ideas for Regulating Generative AInormaltech.ai
- Notes on Creative Transparency - by Elena Burgervirtualelena.co
- AI safety is not a model propertysubstack.com