Can Conversational AI advance the science of talk and language use?

Can the science of language use be advanced using dialog systems, conversational AI, or language models?

Image Credit: Arbor scientiae, by Ramon Lull

Some say that computational modeling is an established method for the investigation of linguistic theory, for testing existing theories and for proposing pausible mechanisms that are involved at all levels of language understanding and generation. In fact, computational modeling using statistical machine learning techniques has become one of the main methodologies in the study of human cognitive processes related to language, despite remaining challenges regarding the interpretability of the outcome of many of such models (Alishahi 2010).

In cognitive science (and other cognitivist research programmes):

Combined with findings coming out of other fields, computational models can give us insights about which representations and processes are more plausible in the light of other empirical and experimental evidence related to the phenomenon under study. The potential contribution of this line of work is twofold, interpreting and evaluating the proposed dialog management models can help to evaluate and advance pragmatic theory, while the model may also be designed with speech technology applications in mind.



Alishahi, A., 2010. Computational modeling of human language acquisition. Synthesis Lectures on Human Language Technologies, 3(1), pp.1-107.

Andreas Liesenfeld
Andreas Liesenfeld
Postdoc in language technology

I am a social scientist working in Conversational AI. My work focuses on understanding how humans interact with voice technologies in the real world. I aspire to produce useful insights for anyone who designs, builds, uses or regulates talking machines. Late naturalist under pressure. “You must collect things for reasons you don’t yet understand.” — Daniel J. Boorstin