Making AI Intelligible by Herman Cappelen & Josh Dever

Author:Herman Cappelen & Josh Dever [Cappelen, Herman & Dever, Josh]
Language: eng
Format: epub
Publisher: Oxford University Press
Published: 2021-03-11T00:00:00+00:00

Schematic AI-Suitable Kripke-Style Metasemantics

In order to de-anthropocentrize the Kripkean story that the analogue of anchoring for AI systems is to be found in their neural net training, very roughly, our proposal is this:

SmartCredit’s outputs express the property of high risk because the training of SmartCredit’s neural network was done against the property of high risk, thereby anchoring the program to that property.

This is the coarse-grained answer, but the details will matter a great deal and they are in large part unsettled. Some of the relevant details involve how neural nets are trained:

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.