Conceptual impressions surrounding this post have yet to be substantiated, corroborate, confirmed of woven into a larger argument, context or network.
* * *
Symbolic Behaviour in Artificial Intelligence Adam Santoro*,1, Andrew Lampinen*,1, Kory W. Mathewson1 , Timothy Lillicrap1 and David Raposo1 *Equal contributions, 1DeepMind 5.
Conclusion
"Classical perspectives on symbols in AI have mostly overlooked the fact that symbols are fundamentally subjective—they depend on an interpreter (or some interpreters) to create a convention of meaning. Thus, human-like symbolic fluency is not guaranteed simply because a system is equipped with classical “symbolic” machinery. Instead, symbolic fluency should be evaluated through behaviours, whether these behaviours involve interactions with interlocutors or simply demonstrate improved internal reasoning. This can be measured by inspecting a set of graded traits, such as receptiveness to new convention, the ability to construct new conventions, and demonstrated understanding of the meaning behind syntactic maneuvers. Because optimizing directly for behaviour is increasingly feasible, we argue that the key to developing machines with human-like symbolic fluency is to optimize learning-based systems for these symbolic behaviours directly by placing artificial agents in situations that require their active use. Human socio-cultural situations are perhaps best suited to fulfill this 9 Symbolic Behaviour in Artificial Intelligence requirement, as they demand the complex coordination of perspectives to agree on arbitrarily-imposed meaning. They can also be collected at scale in conjunction with human feedback, and hence allow the use of powerful contemporary AI tools that mould behaviour."
* * *
* * *
Question: Who is the observer doing the observing while being observed?
No comments:
Post a Comment