Carolyn Meinel's Research Project

AI Governance is critically dependent upon getting all the necessary information.

As Brian LaBatte of BestWorld, the Existential Risks Persuasion Tournament, IARPA FOCUS, and other decision science research projects has observed:

• Does governance evolve or emanate from errors, mistakes, or tragedies, or, from foresight and genius?
• Did the Dodd/Frank governance come from the experience and learning of the financial crisis
• Or ideas waiting to be codified?
• Did the US Constitution come from experience of events in the UK
• Or the philosophies of leading experts?
• Or the combined geniuses of US leaders negotiating and sausage making a constitution?
• Did all of these factors contribute?

Given these many factors, we believe our key contribution could be creating information that is:

• True, including techniques for looking inside the otherwise black boxes of generative AI.
  • Ultimately, our success is critically dependent upon public support for governance measures.
  • DARPA XAI results were discouraging, in my humble opinion.
  • The federated LLMs, humans in several loops architecture of Ai Translate suggests a partial solution.
  • However, my negotiations with them hit a stumbling bloc of an unacceptable NDA.
  • Stay tuned... there are other options.
• Believable, especially for mustering the will of the human race to align AGI.
• Beneficial. The set of "truths" is vast.
  • Decisions needed on what truths to discover and publicize.
  • The AI Alignment community will play a crucial role in such decisions.

 

Home

© 2024 Carolyn Meinel. All rights reserved.