Robin Cohen is conducting research currently aimed at delineating the distinctions between trusted AI systems and the well-established subfield of trust modeling in multiagent systems, a community to which she belongs. Trustable and explanable AI is an important social concern. She has also been involved in research on how to address digital misinformation by applying trust modeling techniques to social media environments. In another recent research effort, she she worked with a group of students from her AI and Philosophy course on a paper presented to the first AI, Society and Ethics conference, examining whether AI Ethics solutions can be provably correct. Other topics relevant to AI and Society covered by Prof. Cohen’s research agenda include effort on advancing assistive technology for cognitively and visually impaired users by depicting webpages as visual images, as well as several explorations of multiagent solutions for the application of AI and Health (improving resource allocation in emergency scenarios, addressing misleading messages in online health forums, reasoning about bother during interaction with medical professionals).

Edith Law is an international leader in human computation and crowdsourcing research, with extensive experience in developing crowdsourcing systems visited by tens of thousands of real users over the web. Her research aims to design algorithms to enable combined human and machine intelligence to tackle complex problems requiring expertise with complex dependencies (including annotation of medical time series and of papyrus documents). She has also promoted the practice of citizen science as a means for overcoming the challenge of automatically analyzing massive sets of data. Her work in this area includes the development of a platform to assist in data collection which also serves as a testbed for studying crowdsourcing questions related to incentives, to task decomposition and to expert-novice interactions. The issue of intrinsic motivational factors to influence worker engagement is especially relevant. This research lies within a current hotbed focus of the AI community, that of perfecting human-AI collaboration.

Kate Larson conducts research related to better understanding how people’s lack of information and bounded rationality can influence voting behaviour and outcomes. She is especially interested in concepts like fairness in computation settings, designing systems which balance incentives of the users with quality of the outcomes. A particular emphasis is on reasoning about the preferences of users, even when people cannot easily articulate them. All of this research is critical for human-AI partnerships to flourish: various roadblocks to making these collaborations successful must be explored and addressed. Prof. Larson has also conducted research for the realworld scenario of wild-fire control, applying computational and game theoretic techniques to resource allocation problems.

Another researcher who has delved into a specific realworld context where AI techniques have the potential to deliver true value is Peter van Beek. He has been working on predicting and forecasting water consumption with a special project for the city of Abbotsford, British Columbia, a location where conservation of water has become a critical concern (with water intake nearing its capacity and with extending the water supply deemed to be extremely costly to consider). He has also been analyzing how the water is being consumed by modeling various restrictions, yielding analysis (along with explanation) in terms of demographic and weather factors.

Jesse Hoey conducts research in affective computing, focusing on understanding social and emotional factors in intelligence, when designing intelligent agents. He is especially interested in exploring social dynamics in online collaborative groups (like github). His research on Bayesian affect control theory has been applied to a host of problems such as social status modeling, human action in a social dilemma and rational decision making for socially aligned agents. Prof. Hoey has also built assistive technologies for persons with cognitive disabilities and in particular for aging, including effort on virtual assistants for elder care: this is AI research with definite social impact.

Jeff Orchard works on neural learning algorithms that are biologically plausible. His theories about how the brain works impact neuroscience and cognitive science, both. This research has the potential for dramatic social impact, informing our understanding of different neuro-cognitive disorders (e.g. Schizophrenia, autism) and how to address them. He is also jointly supervising a Masters student with Prof. Hoey on the topic of building human-like moral decision making into a neural network AI. Prof. Orchard has additional interests in designing neural network architectures that are robust against adversarial or ambiguous input (e.g. misclassification from visual recognition systems). These new directions will improve the lives of those relying on these systems.

Pascal Poupart’s research in machine learning has considered various critical social implications. He has explored in detail assistance for Alzheimer’s patients: through adaptable technology such as sensors, through monitoring with mobile phone GPS and through handwashing reminders suggested from encoding relevant Markov Decision Processes. He has also leveraged computer vision solutions to assist both Alzheimer’s patients and older adults with mobility problems. His work on natural language dialogue has integrated deep learning, affective computing and reasoning under uncertainty, and provides insights into the proper design of conversational agents to assist human users.

Shai Ben-David research has been focused on the safety of ML based systems; ML algorithms that would be able to ”raise a flag” when they encounter a situation that is outside their high confidence regions. He is also building Fairness in ML based decisions. This is especially important as machine learning starts being applied to advice decisions relating to people such as acceptance to university programs where there are growing concerns about the tendency of such systems to use racial, gender or religion as features influencing their decisions. He is also working to improve the interpretability of ML based decisions.

Another researcher with strong interest in investigating how machine learning techniques can serve to make improvements to society is Yaoliang Yu. His focus is on facilitating the transfer of modern advances in machine learning to improve various real life experiences. This is especially important in big data contexts. Prof. Yu is assisting individuals to catalogue their digital life (e.g. from video recordings), through techniques of multiclass support vector machines, as part of knowledge discovery.

Maura R. Grossman has pioneered the use of machine learning, in particular, TechnologyAssisted Review (TAR) and Continuous Active Learning(tm) (CAL(tm)) for electronic discovery in legal proceedings, for curation of public records for government archives, and for systematic review for meta-analysis in evidence-based medicine. Her research has been cited in cases of first impression in the United States, Ireland, the United Kingdom, and Australia. With a background in clinical psychology, hospital administration, law, and computer science, Grossman is especially interested in multidisciplinary approaches to AI. She teaches a unique graduate course that brings together computer science and law students to study the legal, ethical, and policy implications of AI.

Richard Mann has focused on the topic of computational audio. Mann has done previous work in the area of computational vision: on gesture input for human computer interaction, pen-based input systems, and image alignment using Bayesian methods.