It also appears that the link to ELK in this section is incorrect
- Making use of an AI’s internal state,2 not just its outputs. For example, giving positive reinforcement to an AI when it seems likely to be “honest” based on an examination of its internal state (and negative reinforcement when it seems likely not to be). Eliciting Latent Knowledge provides some sketches of how this might look.
The link to ELK in this bullet point is broken.
- It’s not currently clear how to find training procedures that train “giving non-deceptive answers to questions” as opposed to “giving answers to questions that appear non-deceptive to the most sophisticated human arbiters” (more at Eliciting Latent Knowledge).
It may intend to point to here: https://www.alignmentforum.org/posts/qHCDysDnvhteW7kRd/arc-s-first-technical-report-eliciting-latent-knowledge
For more on criterion of rightness vs decision procedure
I don't think my particular VAs have more capacity, but I believe Virtalent has other VAs ready to match with clients.
It is unclear to me whether I’ve just gotten lucky. But with Virtalent you can switch VA and the minimum commitment is very low, which is why I think the best strategy is just to try
I thought the team behind the EAGx designs were really great and I loved them. Have you considered reaching out to them to make designs for your store?