While we moved from command and control to current speech assistants such as Alexa, Bixby, or Siri, these still feel far from fiction such as in Knight Rider, Iron Man, Her, or Ex Machina. Largely unnoticed, however, A.I. is quite further down the road when it comes to Artificial Emotional and Social Intelligence for Spoken Language Interfaces. This concerns in particular the rich already possible analysis of speaker states and traits such as hearing our age, personality, emotion, heart rate, or information on our health state. But “deep” neural networks also already learnt to speak, and to exploit such knowledge in their reaction in a dialogue, for example, to appear more charismatic or empathic.

In this talk, we will discuss the silent transformation of spoken language interfaces towards increased anthropomorphism, and where one can soon expect them to move by upcoming reinforced learning interacting with millions of users at scale from Europe to China and beyond.

Prof. Dr. Björn W. Schuller - FIEEE, CSO - University of Augsburg & Imperial College London & audEERING

Während wir uns an die Kommandos und Kontrolle der Sprachassistenten Alexa oder Siri gewöhnt haben, scheinen die Fiktionen wie bei Iron Man oder Her noch weit entfernt. Gerade bei emotionalen und sozialen Faktoren scheint die künstliche Intelligenz noch weit entfernt vom menschlichen Können. Wie wird sich dies weiterentwickeln?

Veranstaltung im Rahmen des vom BMBF geförderten Projektes CuLTURE China – The KIT Competence Network for Innovation in Cultural Learning and Training in a University and Research Environment.

Homepage
https://www.zak.kit.edu/colloquium_fundamentale.php