June 17, 2024, 16:00 CEST, online (Zoom)

Larry Moss (Mathematics Department, Indiana University, Bloomington): A Place for Logic in the Computer Processing of Language

Starting in 2018, computers have been able to carry out some tasks at human level (or better), tasks which are traditionally thought of as 'logical'.  These include the central task of logic: knowing 'what follows from what', when everything is presented in natural language.   We therefore at a watershed moment in the history of logic.  However, the computational systems -- neural net learners -- do not use logic in any evident manner.  Of course logic is involved in computer science at many levels, but the particular programs involved in inference are much more like the ones that memorize patterns and classify objects.  They do not use explicit symbolic reasoning of the kind logicians love.

Addressed to a general audience rather than to specialists, this talk is concerned with attempts by several groups of researchers to do reasoning in language on the computer, and to probe the deep learners to see how much they really can do, and to create hybrid symbolic/neural reasoning systems.

Program:

16:00-17:00 Lecture by Larry Moss

17:00-17:30 Panel discussion — panelists: Natasha Alechina (Open University & Utrecht), Larry Moss and Bart Verheij (Groningen)

17:30-18:00 Q&A with the audience

Organizers: Dominik Klein (d.klein@uu.nl) and Fan Yang (f.yang@uu.nl)