Dear BU linguistics community,
Mark your calendars for a special talk on January 15!
LiSLab and tinlab looking forward to welcoming the legendary Janet B. Pierrehumbert (Oxford) for a talk on that day. Although she is world-renowned for her work on phonology, she will be presenting recent work on large language models and their relation to formal semantics.
Speaker:
Janet B. Pierrehumbert
Professor of Language Modelling, University of Oxford
Date: January 15, 2026
Time: 11am-noon (followed by lunch)
Place: Room 135 in the linguistics department (111 Cummington Mall)
Title:
LLMs can pass the Turing Test — are they intelligent?
Abstract:
In 1950, Turing proposed that if a person could not tell whether they are in conversation with another person or a computer algorithm, we can consider the algorithm to be intelligent. The Turing Test effectively launched the field of AI, and advanced a strong connection between natural language and intelligence. The newest Large Language Models (LLMs) engage in conversations and produce amazingly human-like output. Often, people cannot reliably tell whether they are talking to a chatbot or another person, and sometimes the chatbot output is found to be better than human output. LLMs seem to pass the Turing Test. Does this mean they are intelligent? Should we worry that AI is on the verge of becoming more intelligent than we are?
In this talk, I will talk about the nature of the Turing Test and the current level of evidence that LLMs can pass it. I will show that LLMs have systematic shortcomings in some areas of language use that reflect intelligence — even in little children. I will trace these shortcomings to aspects of the architecture and training of current LLMs, which cause them to learn inefficiently and fail to capture logical abstractions.
Also check out other Workshops in Allston, Nonprofit events in Allston.
