Sophia Indrajaal
@sophia-indrajaal
184 Following
248 Followers
1 reply
0 recast
4 reactions
1 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
2 reactions
0 reply
0 recast
1 reaction
0 reply
0 recast
0 reaction
1 reply
0 recast
3 reactions
2 replies
0 recast
4 reactions
0 reply
0 recast
2 reactions
1 reply
0 recast
7 reactions
1 reply
0 recast
1 reaction
0 reply
0 recast
1 reaction
1 reply
0 recast
0 reaction
0 reply
0 recast
2 reactions
1 reply
0 recast
7 reactions
1 reply
0 recast
3 reactions
If the map/language structure is a "living" document, first sandbox would need to be around it, a Human-in-the Loop that controls how new entries are added. I have that role now, but in training a larger model rather than a context window instance, this would have to be someone/s outside of the direct DAO, who then needs to be sandboxed from access to other models to train. Lol, I'm not sure any of this is feasible. But the drive is very strong. Wait, no. If the LLM is sandboxed, no internet, no way to communicate to other models that would do it. But how can you build an human/AI DAO of DAOs in a sandbox? Even a proposal could contain enough info to do weird shit. The Sandbox Problem. Answer- if the sandbox is built into the constitution within the map, it should theoretically be stable and self contained in terms of harmful ill effect. Then the problem becomes how to enforce the constitution. This is where a TicTacToe biscuit needs to come in I think. 0 reply
0 recast
0 reaction
0 reply
0 recast
1 reaction
1 reply
0 recast
1 reaction
1 reply
0 recast
1 reaction