We’re excited to announce a new summer outreach program (co-lead by our PI) on combiningAI, Language & Humanities that will take place in Ljubljana, Slovenia.
This program is part of our lab’s commitment to educating students without extensive quantitative background about machine learning, connecting AI with Arts and Humanities to address ethical questions the new technologies bring, and to connecting North American and European higher education.
Arts & Humanities predicted many virtual technologies before they were scientifically implemented and offered the first public space to discuss their ramifications. We will use language and literature to address some of the pressing questions:
How can AI and humanities can inform each other?
How does AI work and how can we use language data to better understand it?
What ethical dilemmas might be anticipated in the future using literature as a testing ground for societal ramifications of new technologies?
The ultimate goal of the program is that students learn basic quantitative skills to understand how AI works, and learn to use arts and humanities as a means to address ethical challenges that the new technologies will bring in the future.
Our lab’s PI was awarded the UW Royalty Research Fund award in the amount of $39,993 to fund the initial stages of a project on computational models of phonetic and phonological learning that uses speech data to better understand how neural networks learn internal representations.
It was a great pleasure to host a visit of CEOs of American Chambers of Commerce in Europe and Honorary Consuls to the Department of Linguistics and UW Phonology Lab to talk about US-Europe academic and industry collaborations, student exchanges, and the current landscape of machine learning and language technologies for underresourced languages.
We trained a Generative Adversarial Network on raw speech data containing evidence for the allophonic distribution of aspiration in English. The network learns to produce an allophonic distribution from raw acoustic data & produces innovative outputs consistent with linguistic behavior.