It-lecture «IT- AI Safety: motivation, open problems, the task of learning human values»

Creating human-level AI: Will it happen, and if so, when and how? What are the remaining key obstacles? What can we do now to maximize the probability of a positive outcome? First we will examine the motivations behind the field of AI Safety (AIS), followed by a discussion of the main strategic considerations for AIS research. Next, we will explore the main areas of AIS research. Finally, we will delve into the problem of learning human values and introduce the Cooperative Inverse Reinforcement Learning framework (Hadfield-Menell et al. 2016), an insightful formalization of the value alignment problem. 

The event will be held in English, the speaker is Dmitry Krasheninnikov graduate of South Eastern Finland University of Applied Sciences, Master of Arts in Artificial Intelligence at the University of Amsterdam (since September 2017), visiting researcher at Center for Human-Compatible AI, UC Berkeley.

To register for the event, please fill out the form.