532C / 554C (cross-listed)

Human-Centered AI
Fall  2019
Cristina Conati

Class Data Course Description Coursework Sample Contents Schedule and Readings Marking Scheme  

 Class data
Time: Th.  2pm - 5:00 pm.
Location: ICICS/CS 246
Instructor Office: ICICS/CS 107
Office Hours: By appointment
Email: conati@cs.ubc.ca
Piazza Register at

piazza.com/ubc.ca/winterterm12019/cpsc532c554c

Course mailing list cpsc532c@cs.ubc.ca 
Subscribe  by sending the message "subscribe cpsc532c" to Majordomo@cs.ubc.ca.
   
Course description

This course explores  how to design and evaluate  AI-driven, human-centered interactive  systems capable of both performing useful tasks as well as  being well accepted by their users. A key aspect of this endeavour is enabling AI systems to predict and monitor relevant properties of their users (e.g., states, skills, needs) and personalize the interaction accordingly, in a manner that maximizes both task performance as well user satisfaction, abiding to principles  of transparency, interpretability, predictability and   user-control


Tackling these challenges requires an interdisciplinary effort that integrates research in different areas of Artificial Intelligence (e.g., machine learning, reasoning under uncertainty, natural language processing, planning, Explainable AI) with research in Human  Computer Interaction, and  Cognitive Science. During the course we will explore representative work in this field, spanning the above disciplines



Coursework
(click  on each activity below for more information)


Readings

Most classes will be devoted to the discussion of a selection of research papers, which will be posted on the course reading list (to be posted) Students are required to read the papers in advance.

Summary/Questions on the readings

Discussion on the assigned readings will be the main activity in the course. To provide good ground for the discussion, participants must

  • send   at least two questions and a brief summary as specified on the course reading list (to be posted later).  

  • part of the summary/questions for each class will be due by the morning of   Monday before class, the rest  will be due by Wednesday evening before class.  

    • Details on which material to send, and when, will be posted on the course reading list. 

    • Material sent after the deadline will be marked as zero. However

    • Each student has 2 "no paper" bonuses can avoid  sending the material for 2 papers with no penalty.

  • Clarification questions are welcome. However, there should be  at least two   questions on each paper  that do one or more of the following:

    • address weaknesses in the presented research

    • relate the research to general issues in the field

    •  make connections/comparisons with other readings

    • here are  some sample questions

  •  Each paper summary should be based on the following template and be no more than 2 pages long. As shown in the template , the summary should include 4 headers addressing  the following points, described in more details here

    • What are motivations for this work?
    • What is the proposed solution?
    • Has the proposed solution been evaluated, and if so how ?
    • What are the contributions of this work?

 

Beside improving  participation to the discussion, the objective of these activities  is to help participants learn how to read papers with a critical eye, as a reviewer for a conference or journal would do.


Paper presentation/discussion leading

One or more  times during the term (depending on class size), each participant will present  a  reading for  that class and will lead the discussion on it

Paper presentation involves preparing a few slides with a critical summary of each assigned paper, including

  • the same 4 points to be covered in a regular paper summary
  •  a categorization of  each paper/system following  schemas  from the intro class

This part should take at most 10'. If you have more than 10-12 slides, your presentation is likely too long. Time your presentation to make sure that it is of the right length.

Discussion leading will  include

  • collecting the questions posed by the rest of the class
  • structuring  the discussion around (a selection of) these questions
  • proposing answers
  • having questions/discussion points ready to stimulate the discussion

Presenters do  not need to send summaries/questions on their assigned papers

Project

  • Participants will  be expected to complete a course project. The project will be decided in consultation with the instructor and will typically involve implementing a simple adaptive system.

  • Projects can be done in teams of two students

  • Each student/team  is  expected to set up individual meetings with the instructor to discuss project options. The earlier this is done, the easier it will be  to come up with a suitable proposal, but plan to schedule at least one meeting before  mid February (date TBA)  

  • The project will include the following stages:

    • a project proposal (max. 3 pages the following format) ), (due  date TBD) which will be presented during that day's class

    • presentation on project progress toward the end  of the course (date TBD).  The project  should be  in an advanced state of completion  by this date.

    • Project completion and  written report (max. 20 pages using  the following format) (due date TBA)

     

    For a team project,  the proposal must clearly define  the contribution of  each group member. Each group member must present his/her contribution during project presentations

  • GUIDELINES for selecting a  project