"LazImpa": Lazy and Impatient neural agents learn to communicate efficiently
This addresses the inefficiency in emergent communication for AI agents, offering a solution to align artificial codes with natural language patterns, though it is incremental as it builds on prior referential game setups.
The paper tackled the problem of inefficient communication codes in neural agents, showing that near-optimal and Zipf Law of Abbreviation-compatible messages emerge when both speaker and listener are modified with laziness and impatience mechanisms.
Previous work has shown that artificial neural agents naturally develop surprisingly non-efficient codes. This is illustrated by the fact that in a referential game involving a speaker and a listener neural networks optimizing accurate transmission over a discrete channel, the emergent messages fail to achieve an optimal length. Furthermore, frequent messages tend to be longer than infrequent ones, a pattern contrary to the Zipf Law of Abbreviation (ZLA) observed in all natural languages. Here, we show that near-optimal and ZLA-compatible messages can emerge, but only if both the speaker and the listener are modified. We hence introduce a new communication system, "LazImpa", where the speaker is made increasingly lazy, i.e. avoids long messages, and the listener impatient, i.e.,~seeks to guess the intended content as soon as possible.