In this paper we discuss how Robert Brandom’s inferentialist project (as discussed in Between Saying and Doing), illuminates importante aspects of contemporary debates about Explainable Artificial Intelligence (XAI).
Abstract
Classical Artificial Intelligence has a foundational place in Brandom: the practices whose domain constitutes the possession of a vocabulary are the application of a series of algorithms. Making explicit these algorithms provi-des an explanation for Brandom’s project of bringing the inferential commitments implicit in our practices into the game of giving and receiving reasons. This project fails for a reason well known in AI, the frame problem. Brandom proposes a solution to the frame problem through learning by training. Brandom’s proposal comes close to neural networks developed through machine learning. While this approach does not allow us to maintain the Brandomian framework of Between Saying and Doing, the parallel with Making it Explicit brings an important parallel with the project of Explainable Artificial Intelligence, namely making explicit the implicit inferential commitments of decision-making processes that affect our common life.
Authors
- Ernesto Perini-Santos
- Carlos Barth
Idiom
Português
DOI
https://doi.org/10.26512/rfmc.v13i2.55078