WebEx Link: https://fu-berlin.webex.com/fu-berlin/j.php?MTID=ma8d91ff0fabfcc30e8b464e671a12230

Description

Over the course of the last decade, machine learning has transformed not only AI as a field, but also many sectors of the economy. It is being successfully applied in predictive maintenance, smart grids, and vaccine development. Yet, ML has its limits. ML agents cannot be imaginative. They cannot perform well without a large quantity of high-quality data. And they cannot infer additional information from the context in which they are deployed.

In this seminar students will explore the practical limits of machine learning and venture into other AI approaches that are informed by the scientific study of the mind. Thereby, they will learn about the multifaceted nature of human intelligence. And they will scrutinize how AI developers design agents that exhibit human behavioral patterns in the context of multi-agent simulations, video games, and humanoid robots.

By the end of the course, students will gain a better understanding of the challenges associated with moving from narrow AI to general artificial intelligence. They will also appreciate why the pursuit of general AI is an integral part of the endeavor of advancing artificial intelligence.

Practical Aspects

• Each student will choose a topic. Students will have to study the literature, prepare a presentation, and write an essay.
• Essays must be submitted by the end of the semester.
• The lecturer will give the first three lectures. Afterwards, each class will consist of a student presentation and an extended discussion. Students are also expected to participate actively in the discussion.
• The evaluation depends on three factors: (1) presentation; (2) written essay; (3) participation in class discussion.

Prerequisites

This course is ideally suited for Master’s students. Having attended an introductory course to AI would be highly beneficial.

Instructor

Dr. Nabil Alsabah holds a PhD in psychology and a Master’s degree in computer science. In the past, he spent research stays at Peking University, National University of Singapore and Stanford University. The lecturer looks back at a 15-years experience researching and working at the intersection of artificial intelligence and psychology.

Topics

• The dream of artificial general intelligence: Why did AGI permeate the early decades of AI history? How did the founders of AI envisage the future of the field? Why did the dream fade? Literature: McCorduck, 2019. => Karim Ismail (09.11.2021)
• Limits of machine learning: What areas are unsuitable for applying machine learning? What alternatives are being pursued at research institutes? What are their pros and cons? => Fang Lin (16.11.2021), Darius Golagha (23.11.2021)
• Cognitive models: What are cognitive models? How are they structured? Who develops them? What real-life applications are there for cognitive models? Literature: Bermúdez, 2020; Ritter et al., 2019. ==> Marisa Nest, (30.11.2021)
• Agent behavior in multi-agent simulations: What are some typical multi-agent simulations where agents have to exhibit human-like behavior? How do agents usually behave in such scenarios? How do designers address the limitations of AI in this context? Literature: Jezic et al., 2016; Sun, 2008. => Gianluca Volkmer, (07.12.2021)
• AI of NPCs in video games: How are NPCs usually designed? What intelligent behavior do they generally exhibit? How does AI feature in NPC design? Literature: Kopel and Hajas, 2018; Merrick, 2008. => Franziska Rau, (14.12.2021)
• Environmental design of multi-agent simulations: What characterizes realistic virtual environments? What are the principles for designing realistic multi-agent scenarios? How do designers go about designing a virtual environment? Literature: Dörner, 1997. Michael Hoffmann, (04.01.2022)
• Virtual environments as playground for advancing AI: How does the environment influence agent behavior? What kind of knowledge do AI researchers need to design authentic virtual environments? What examples are there for realistic virtual environments? Literature: Neo et al., 2021; Schell, 2019. => Numan, (11.01.2022)
• Imbuing humanoid robots with human-like behavior: Where do we see humanoid robots? What behavioral patterns do they exhibit? What are the technical challenges for making robots behave like humans? Literature: Gelin and Laumond, 2018. => Jacob Jacobfreise, (18.01.2022)
• Centrality of language for achieving human-level intelligence: Why is language understanding a crucial component of intelligence? Why is it difficult for AI to understand language? Where does modern AI stand regarding language? Literature: Russell and Norvig, 2019 (Chapter 24). => Friedrich Müller (25.01.2022), Omar Hussein (01.02.2022)
• Ethical questions related to AGI: What are the most important ethical questions being discussed in the context of AGI? What worries people the most? How to address those worries? Literature: Russell, 2020. => Jonas Schäfer (08.02.2022), Nikita Naumov, (15.02.2022)

Literature

Bermúdez, J.L. (2020) Cognitive Science. Cambridge University Press, New York.

Dörner, D. (1997) The Logic Of Failure. Perseus, Cambridge, MA.

Gelin, R. & Laumond, J.-P. (2018) Humanoid Robot Applications: Introduction. In Humanoid Robotics: A Reference, Springer Netherlands, Dordrecht, pp. 1-4.

Jezic, G., Howlett, R.J. & Jain, L.C. (2016) Agent and Multi-Agent Systems: Technologies and Applications. Springer,

Kopel, M. & Hajas, T. (2018) Implementing AI for Non-player Characters in 3D Video Games. In Intelligent Information and Database Systems: Lecture Notes in Computer Science, Springer International Publishing, Cham, pp. 610-619.

Marcus, G. & Davis, E. (2020) Rebooting AI. Vintage, New York.

McCorduck, P. (2019) Machines Who Think. A K PETERS, Natick.

Merrick, K. (2008) Modeling motivation for adaptive nonplayer characters in dynamic computer game worlds. Computers in Entertainment, 5, 1-32.

Neo, J.R.J., Won, A.S. & Shepley, M.M. (2021) Designing Immersive Virtual Environments for Human Behavior Research. Frontiers in Virtual Reality, 2,

Ritter, F.E., Tehranchi, F. & Oury, J.D. (2019) ACT-R: A cognitive architecture for modeling cognition. Wiley Interdisciplinary Reviews: Cognitive Science, 10, e1488.

Russell, S. (2020) Human Compatible. Penguin Books, New York.

Russell, S. & Norvig, P. (2019) Artificial Intelligence. Pearson Higher Education, Hoboken.

Schell, J. (2019) The Art of Game Design. CRC Press, Boca Raton.

Sun, R. (2008) Cognition and Multi-Agent Interaction. Cambridge University Press, New York.