A recorded livestream discussion on ethical issues in generative artificial intelligence featuring CJ Trowbridge, Future Inifinitive, and The Real Cornpop.
My main takeaways of the topics discussed:
- The practical impossibility of any real ban or limits in a post-international-law world
- Lack of explicit consent vs fair use rights
- Derivative work vs stolen work
- Alignment
- Elon assumes he can force AI to act like Stephen Miller; it’s more likely to act like the hundreds of thousands of administrative beurocrats who refused to go along with him.
- Consolidation of power/ wealth
- Epistemic Power
- Centralized corporate/oligarch control over the social construction of truth and knowledge
- The only truth is whatever serves capital
- Bias and representation
- What perspectives does a model built by {Sama, Elon, Zucc} ignore, exclude, or erase
- Cloud Capital
- Very large models are almost exclusively built/run by rent-seeking coprorations/oligarchs
- Epistemic Power