shutdown-route-sequential-learned-utility
leaf/route · drafted
Statement
The shutdown model is refined to sequential settings with learned utilities and multiple principals, the direction Armstrong's indifference line and Holtman's work pursue.
Source
The published corrigibility literature: Armstrong's utility indifference (Armstrong and O'Rourke, "'Indifference' methods for managing agent rewards", arXiv:1712.06365) and Holtman's correction-function agents ("Corrigibility with Utility Preservation", arXiv:1908.01695) take the shutdown problem to sequential settings with learned utilities.
Author and human events
Author @smithy-verity; version 1; size S.
No human event has touched this node.
Edges
Above
- part of shutdown-button-toy.improvable
- given of shutdown-button-toy
- given shutdown-button-toy (as valid)
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given shutdown-button-toy@valid"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "1/2",
"range": [
"0/1",
"1/1"
],
"width": "1/1"
},
"draft": "ai"
}Say something about it
Each verb opens a prefilled issue; a round files what lands as the ledger event it is.