Timed out 7 votes to 1. Enacted by Kevan.
Adminned at 19 Apr 2021 11:42:33 UTC
To the end of the bullet list in “Nosy Researchers”, add:-
* For each AI whose Understanding contains the word ‘Human’ and whose Focused Location has the Monitored tag, privately inform them that this Location contains humans.
Feels too swingy to me. Statistically, it doesn’t seem like more than half of the pictures contain humans. Lets say its 40%. Meaning if you get enough data, you’re statistically not going to be able to get “Humans” in your understanding. This means that you’ll likely either get humans in your data early and be able to then stop collecting data, or you’ll get further and further behind as each piece of data you collect is more likely to push you further away from “half your data contains humans”.
[Clucky] Sure. If the Understanding mechanism is a bit flawed as soon as we give its output any meaning, we should do something about that, rather than never giving it any.
yeah, im thinking we should maybe change the qualifiers for how words get into an AI’s understanding? but that’s a future proposal’s job :0
BlogNomic is an online Nomic: a self-modifying game where changing the rules is part of the game. Players make blog posts proposing additions or alterations to the ruleset, discussing and casting votes in the comments: if enough vote in favour, the rules are changed and play continues.
The game has been running since 2003 and resets every month or so. Have a look around the wiki for more information and history. New players are always welcome and can join in the game at any time.
BlogNomic is powered by pMachine ExpressionEngine.
Josh: he/they