Case Teaching Notes
for
The “Living” Room: A Case Study in Artificial Intelligence, Collaborative Systems, and Language Understanding

by
Stephanie E. August
Department of Electrical Engineering and Computer Science
Loyola Marymount University Los Angeles

Introduction / Background

This case analyzes the reasoning processes and types of information that need to be embedded in collaborative software systems in order for these systems to demonstrate intelligent behavior and allow people to interact with them in a natural way. It is based on, and quotes sections from, the Overview of M.I.T.’s Project Oxygen: Pervasive Human-Centered Computing found on the M.I.T. Project Oxygen website [1]. In the case, a college student named Kate carries on a conversation with her dorm room, which she has named “Alice.” Alice collaborates with Kate to adjust the room temperature when Kate complains about it and, recalling a previous conversation with Kate, suggests that she join her friends for a trip to Disneyland to cure her boredom.

The March 2001 issue of Communications of the ACM [2] was devoted to projections of how we will experience technology over the next thousand years. Clothing will adapt to its wearer and monitor our health [1]. Software will learn our patterns of behavior and wisely anticipate our needs just as Radar O’Reilly anticipated Colonel Potter’s needs in the movie M*A*S*H [3]. Our software applications will no longer crash and our devices will do a far better job of detecting and resolving problems within themselves. We will be able to carry on conversations with our refrigerators, washing machines, and room, interacting with them using speech, text, and gestures [4, 5, 6, 7, 8]. Systems of the future will not simply act as our servants [9]. They will be able to understand what we are trying to accomplish and to collaborate with us to achieve our goals [10, 11].

While most students enjoy movies and novels that describe such a future, many do not believe it is possible, or have no idea how it will be achieved. This case challenges students to imagine what it will take to achieve this future in which our environment is a collaborator and constant supportive companion.

Although the case is designed for computer science students in an introductory course in artificial intelligence, it has broad applicability. With a different set of questions, it could be re-cast to focus on critical thinking skills and be used by non-majors. Students with little technology background, such as those enrolled in Computer Science for Non-Majors to satisfy a general science requirement, who complete this case will become aware of the current and future capabilities of the systems with which they interact and the ubiquitous, embedded nature of computers in the future. They will also come to realize the ways in which a computer can display emotion, regardless of whether it feels emotion [12, 13]. At the other end of the spectrum, students with some computer science background, such as a junior computer science major who has taken courses in data structures and operating systems, can apply their knowledge to engineer a system that will exhibit the behavior described in the case. Students who are taking a course in artificial intelligence and have some exposure to cognitive modeling and natural language understanding can develop computational models of the behavior of the machines and software, as well as lexical entries and procedures for parsing the words of a dialog. The application of the case to these different student audiences can easily be accomplished by carefully tailoring the questions which students must address.

In addition to reinforcing principles of computer science and artificial intelligence, the case is intended to enhance computer science students’ abilities to discuss their work with people who are not engineers. The field needs people who are capable communicators. By discussing the ideas behind the technology independent of their implementation, the case reinforces the student’s ability to share technology with people in other fields.

As a general note, while there are numerous ethical issues that arise when machines become collaborators, aside from touching on those issues in the case in the negotiations between Kate and Alice surrounding the temperature of Kate’s dorm room, ethical issues are beyond the scope of this case study.

The case was originally presented at the Case Study Teaching in Science Workshop at the University at Buffalo in May 2006 and subsequently demonstrated to faculty at the Center for Teaching Excellence at Loyola Marymount University, Los Angeles, in October 2006. It has been used in an undergraduate Introduction to Computer Science for Non-majors course and a senior-level Artificial Intelligence course for computer science majors.

Objectives

The case is designed to introduce students to the knowledge and reasoning capabilities that we will expect in devices and applications of the future, and to demonstrate that these systems will rely on existing technologies. They should understand the concept of a collaborative system and understand that collaborative systems utilize existing algorithms yet provide additional services or functionality.

Upon completing the case, students should be able to:

• Map words to a smaller set of primitives or to the ideas intended behind the words.
• Make causal connections.
• Recognize scripts in text.
• Identify goals and plans.
• Analyze a speech act.
• Understand how expectations guide processing.
• Identify ways in which a machine or software program can respond that a human would interpret as a display of emotion.
• Recognize instances of collaboration in software and interactive devices.
• Identify some of the issues related to human/machine collaboration.

Classroom Management / Blocks of Analysis

Students work on the case in groups of three to four students per group. The case can be presented during a 75-minute class meeting using the following timeline and five to seven of the questions included with the case:

• 5 minutes to organize into groups and hand out the case.
• 10 minutes to introduce the case.
• 35 minutes for group discussion.
• 20 minutes for reporting and discussing group results.
• 5 minutes for concluding remarks and presentation of the follow-on assignment.

If the class has a large number of groups, consider using the last few minutes of the group discussion time for group scribes to pair up and combine their reports, so that the subsequent number of groups reporting is cut by half.

Student Preparation

Prior to introducing the case, students should be introduced to the general concept of artificial intelligence (AI). This would include making students aware that there are two approaches to AI. The first is that AI is like dissecting a crawfish or frog in biology, what computer scientists refer to as the “white box approach.” We want to ask the questions, “What’s inside? How does it work?” and develop cognitive models of human problem solving and reasoning. The second approach is to consider intelligence to be a “black box.” We say, “Who cares how it works? I just want it to act smart!” and develop systems that demonstrate behavior that would generally be considered intelligent, without regard for how it is implemented.

Students in a senior-level artificial intelligence course should also have some background preparation in the cognitive aspects of natural language understanding. One view of natural language understanding is as follows: At the sentence level, words can be mapped to conceptual dependencies [20], a constrained set of primitive concepts. Understanding can be viewed as the process of creating a map of the concepts being conveyed by the words in the sentences and capturing the causal connections between sentences. The causal chains map to scripts, which are stereotyped sequences of actions that define prototypical situations. If a script is recognized, it can trigger expectations for other information that might be represented in the text. Associated with each script are roles that the actors assume in the script, tracks or variations in the way the script can play out, and scenes that commonly occur in the tracks.

Thus a restaurant script might have customer and waiter roles, tracks for cafeteria or sit-down restaurant, and order and pay scenes. A goal, such as satisfy hunger, is an objective or state of affairs that is achieved by finding or devising a plan and carrying it out. A plan, such as buy food and cook it or go to a restaurant, consists of a series of predetermined steps to achieve a goal and can be formulated using familiar scripts.

The words of a text help the understander (or reader) recognize the script reflected in the text, which in turn helps the understander identify the plan and then the goal in the ideas being conveyed. Likewise, if the reader is told the goal, the reader can anticipate which plans might be relevant and which scripts will achieve the plans. In this way, scripts, plans, and goals provide expectations for what the reader will encounter as well as a framework to use in inferring characters’ actions or events that are not explicitly related in the text.

Many students have been introduced to the idea of concept maps in middle school. The instructor can build upon this to introduce the idea that natural language is a way of packaging a concept map or set of ideas. Another person (or computer) receives a package containing a set of sentences, unwraps the package, so to speak, and maps the words of the package to the concepts and beliefs that the receiver holds.

Student preparation involves exploring related resources on the web and preparing short answers to a few discussion questions.

Case Presentation

Begin class with a brief discussion of the elevator statements. Distribute the case and instruct the groups to designate team members to act as time-keeper (to ensure that the team completes all tasks in the allotted time), discussion chair (to keep the conversation moving and cover all topics, as well as eliciting participation by all team members in the discussion), secretary (to record the discussion), and reporter (to share the results with the class following the group work). Let them know that at the end of the discussion each group will need to turn in their answers to the case questions. Individual students might want to keep their own notes as well, to use when completing the subsequent homework assignment.

Each group should spend a few minutes reading the case, then sharing the results of their pre-case research before discussing the 11 questions associated with the case (answers to these questions are provided in the Answer Key that accompanies this case). Encourage students to reference the background readings in their responses where relevant.

Fifteen minutes before the end of class, allow each group to share their response to one or two questions (depending upon class size) and elicit reactions from the rest of the groups. Present the following questions to round out the discussion:

1. Do you think that collaborative systems are possible now? In 5 years? Ever?
2. Can a computer carry on a meaningful conversation or dialog with a person? If so, what is needed to make this possible? If not, why not?
3. Can a computer display emotion through its interface(s)?

Just before class ends, tie together the points discussed by the students. Possible concluding thoughts:

A machine does not have to have emotions to display behavior that the user will interpret as conveying a particular emotion. If it interacts with the user in ways that seem human, the human can infer emotion from the computer’s behavior or response, regardless of whether the computer actually experiences those emotions.

Machines of the future will need to know much about human needs and behavior in order to interact with and assist humans in a natural way.

Written Research Assignment

At the end of class, ask the students to write individual answers to the following questions as a follow-up assignment:

1. What is collaboration?
2. What is a collaborative system?
3. List some real world examples of collaborative systems that you have read or heard about. Describe the task accomplished and the form the collaboration takes.
4. What products (hardware and/or software) that you currently use demonstrate the ability to collaborate? [Answers might include PDA (e.g., Blackberry, Palm) calendar software used to enter goals, tasks priorities, plans and provide schedule and a daily to-do lost which can be reprioritized for the short term…]
5. What is required in order to carry on a meaningful dialog rather than the shallow conversation that one might have with a chatbot? [Many aspects of communication can be mentioned here: context, domain knowledge, reasoning ability, ability to remember what was recently communicated, reasonably sized vocabulary, ability to recognize sentence and discourse structure and to generate language in line with those structures, ability to generate idiomatic expressions. The idea is for the student to give the question some thought, rather than fully enumerate what is needed.]
6. What task would you like a computer to help you perform?

Possible Follow-up Assignments

• Students in an AI class can be tasked with developing a computational model of the objects and rules needed to infer that a trip to Disneyland is possible, given a bored individual with a friend who has extra tickets and a car. The inference process can be modeled using a tool such as CLIPS [23].
• Students in an AI class with access to a phrasal parser can be tasked with developing the logic, phrases and functions needed to incorporate “incidentally” into their lexicon.

Exam Questions Covered in the Course of the Case Study

1. What is a collaborative system?
2. What is required to understand natural language?

Answer Key

Answers to the questions posed in the case study are provided in a separate answer key to the case. Those answers are password-protected. To access the answers for this case, go to the key. You will be prompted for a username and password. If you have not yet registered with us, you can see whether you are eligible for an account by reviewing our password policy and then apply online or write to answerkey@sciencecases.org.

References and Resources

[1] MIT Project Oxygen: Pervasive, human-centered computing.

http://www.oxygen.lcs.mit.edu/

[2] Communications of the ACM. 44:3, ACM Press, NY, March 2001.

[3] Defense Advanced Research Projects Agency, Information Processing technology Office, Personalized Assistant that Learns.

http://www.darpa.mil/IPTO/programs/pal/pal.asp

[4] COLLAGEN: Java Middleware for Collaborative Agents.

http://www.merl.com/projects/collagen/

[5] DiamondHelp: Collaborative Help for Networked Home Products

http://www.merl.com/projects/diamondhelp/

[6] FormsTalk: Multimodal Mixed-Initiative Form Filling.

http://www.merl.com/projects/FormsTalk/

[7] Rich, C., and C.L. Sidner. 1998. COLLAGEN: a collaboration manager for software interface agents. An International Journal: User Modeling and User-Adapted Interaction 8(3/4): 315-350 (Kluwer Online, TR1997-21a).

[8] Bickmore, Timothy W., Lisa Caruso, Kerri Clough-Gorr, and Tim Heern. 2005. “It’s just like you talk to a friend”—relational agents for older adults. Interacting with Computers 17(6): 711–735.

[9] Powell, Alvin. 2002. AI evolution: From tool to partner. Harvard University Gazette January 31, 2002.

http://www.news.harvard.edu/gazette/2002/01.31/10-grosz.html

[10] Forbus, Kenneth D., and Thomas R. Hinrichs. 2006. Companion cognitive systems: a step toward human-level AI. AI Magazine [American Association for Artificial Intelligence] 27(2): 83–95.

https://www.aaai.org/ojs/index.php/aimagazine/article/view/1882/1780

[11] Human Robot Interaction for Hosting Activities.

http://www.merl.com/projects/hosting

[12] [Kismet] Emotions: Emotions in Living Systems

http://www.ai.mit.edu/projects/sociable/emotions.html

[13] Kismet Home

http://www.ai.mit.edu/projects/humanoid-robotics-group/kismet/kismet.html

[14] 2001: A Space Odyssey (film). Directed and produced by Stanley Kubrick. Screenplay by Stanley Kubrick and Arthur C. Clarke. 1968.

[15] Stork, David G. HAL’s Legacy: 2001’s Computer as Dream and Reality. M.I.T. Press, 1998.

[16] Colossus: The Forbin Project (film). Directed by Joseph Sargent. Written by James Bridges based on the novel Colossus by D.F. Jones. Produced by Stanley Chase. 1970.

[17] Jones, Dennis Feltham. Colossus. Macmillan, 1968.

[18] I, Robot (film). Directed by Alex Proyas. Written by Jeff Vintar and Akiva Goldsman based on the story Runaround by Isaac Asimov. (2004)

[19] Asimov, Isaac. Runaround. Originally published in 1942, reprinted in I, Robot. Grafton Books, London, 1968.

[20] Schank, Roger C., and Robert P. Abelson. Scripts, Plans, Goals, and Understanding: An Inquiry Into Human Knowledge Structures. Hillsdale, NJ, L. Erlbaum Assoc., 1977.

[21] Natural Language Processing.

http://www.aaai.org/AITopics/html/natlang.html

[22] Podevin, Jean-François. Various works of art.

http://www.pelavin.com/-podevin.html

[23] CLIPS: A Tool for Building Expert Systems.

http://www.ghg.net/clips/CLIPS.html

Acknowledgements: This case was developed with support from the National Science Foundation under CCLI Award #0341279. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Date Posted: September 05, 2008.

Copyright © 1999–2009 by the National Center for Case Study Teaching in Science.  Please see our usage guidelines, which outline our policy concerning permissible reproduction of this work.