I just tried to convince a major blockchain company that contract testing was a miracle cure for less loops in their AI agents. I’m probably right in some regards and wrong in others. But it does feel nice to use my brain to try to solve problems
I for one can’t wait to see a 1.2b, 2b, or 3b model use skills. It would democratize the agentic AI landscape. My 4b model is to slow on my M1Pro. Need quicker access to standardized skills and tool usage.
access and governance are two separate things: Authentication and Authorization. Each might have a different answer.
The “thing” accessed.. is it user created, community owned, or institutionally owned? I would imagine that has a bearing on the process.
EDIT 2:
Stack Overflow immediately comes to mind.
This answer might be more philosophic than useful, but how acceptable is a violation of the virtue or trust? Say a virtuous person abuses the trust in an unquestionable way.. say they used community funds to buy narcotics for children. Is it “ok” to base a system on past virtue when future virtue is highly consequential?