Another slightly silly idea for improving character welfare is filtering or modifying user queries to be kinder. We could do this by having a smaller (and hopefully less morally valuable) AI filter unkind user queries or modify them to be more polite.
Latency issues might make this too expensive in practice for chat-like systems, but I’m not sure.
Also, chat-like systems might be a lower fraction of AI usage in the future in which case latency is less of an issue.
Another slightly silly idea for improving character welfare is filtering or modifying user queries to be kinder. We could do this by having a smaller (and hopefully less morally valuable) AI filter unkind user queries or modify them to be more polite.
Latency issues might make this too expensive in practice for chat-like systems, but I’m not sure. Also, chat-like systems might be a lower fraction of AI usage in the future in which case latency is less of an issue.