Exploring, Assisting, and Improving Human Rationality using Computational Approaches
bachelor thesis (2022)
Status | in progress |
Student | Shiyi Gou |
Advisor | Changkun Ou |
Professor | Prof. Dr. Andreas Butz |
Period | 2022/05/20 - 2022/08/20 |
Task
Description
Recent advances in AI algorithmic research well supports machine intelligence in human System 1 tasks such as image classification, segmentation, captioning and etc. However, it is also descriptively discovered in psychology that the human mind is not only computational, bounded rational but also adaptive. These reasons make machine intelligence often fail to be used when involving human actions, such as preference choices, due to the violation of algorithm assumptions and result in unsatisfactory outcomes. For example, in 3D modeling, there are rich assistive visualizations to support user actions and their modeling process. Still, the eventual outcome quality is bound to the human-perceivable differences of the software outcomes, user expertise, and the reliability of a decision when choosing different design alternatives. The thesis aims to explore human limits regarding their decision rationality in 3D graphics or relevant fields. Then proposing novel designs and develop systems/mechanisms to assist and improve these limitations and their behaviors based on the verifications of user studies.
Requirements
- Understanding HCI research approaches
- General knowledge about machine learning
- Interests in cognitive science, and also align with behaviorism and connectionism
- Capable and enjoy coding
Initial Readings
- Lewis, Richard L., Andrew Howes, and Satinder Singh. "Computational rationality: Linking mechanism and behavior through bounded utility maximization." Topics in cognitive science 6.2 (2014): 279-311.
- Gershman, Samuel J., Eric J. Horvitz, and Joshua B. Tenenbaum. "Computational rationality: A converging paradigm for intelligence in brains, minds, and machines." Science 349.6245 (2015): 273-278.
- Antti Kangasraasio, Kumaripaba Athukorala, Andrew Howes, Jukka Corander, Samuel Kaski, and Antti Oulasvirta. 2017. Inferring Cognitive Models from Data using Approximate Bayesian Computation. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI '17). Association for Computing Machinery, New York, NY, USA.