The language learning process of a child begins with the babbling and repeating of syllables and goes through various stages of child development. Language is a highly complex world of communication, but we still know a lot about it. That’s why language in Beguš’s lab serves as a starting point for exploring what goes on in the black box of neural networks.

Deciphering how artificial intelligence models learn is mostly a big unknown. A better understanding of neural networks is possible by introducing linguistic approaches. Tune in for the first part of the podcast Language Conversations on the Ars Program (in Slovenian), which is about “talking artificial intelligence” and the creative aspect of language in everyday human communication.

Gašper Beguš is an ASEF professor and holds a PhD from Harvard University. Prior to his high-profile move to the University of California Berkeley, he was a professor at the University of Washington, where he successfully directed the Phonology Lab.

More about artificial intelligence and language (a lecture in Slovenian).
And Why phonology/speech science is so essential for Deep Learning and cognitive modeling (a talk in English).

Učenje jezika se pri otrocih začne z bebljanjem, ponavljanem zlogov in poteka skozi različne razvojne stopnje. Jezik je izjemno kompleksen komunikacijski svet, a o njem vseeno vemo veliko. Prav zato v Beguševem laboratoriju jezik služi kot izhodišče za izvajanje raziskav na temo učenja nevronskih mrež. Razvozlavanje, kako se modeli umetne inteligence učijo, je namreč ena velika neznanka.

Prav z vpeljavo jezikoslovnih pristopov pa je možno boljše razumevanje delovanja nevronskih mrež. Prav z vpeljavo jezikoslovnih pristopov pa je možno boljše razumevanje delovanja nevronskih mrež. Vabljeni k poslušanju prvega dela oddaje Jezikovni pogovori na Programu Ars, ki naslavlja govorečo umetno inteligence in kreativni aspekt govora tekom vsakodnevne človeške komunikacije.

Gašper Beguš, Ph.D. je ASEF profesor, ki je doktoriral na Univerzi Harvard. Pred odmevno selitvijo na Univerzo v Kaliforniji, Berkeley, je bil profesor na Univerzi Washington, kjer je tudi uspešno vodil fonološki laboratorij.

Več o umetni inteligenci in govoru (v slovenščini) (predavanje v slovenščini).
Zakaj so fonologija in govorne znanosti pomembne za globoko strojno učenje in kognitivno modeliranje (nastop v angleščini).