Thought experiments as investigation tools on value conflicts in application of artificial intelligence in education (AIED)

• In Event: Technology and education. Inclusive education for domestic and international students

Proposal

The emergence of a new technology often obscures its social implications. In the long run, the repercussions of ethical issues may slow down its development process. Many schools have already adopted Artificial Intelligence (AI) to support teaching, learning and management. However, the developer and user tend to neglect the value demands of stakeholders in the educational scenario. AI in education (AIED), such as ghostwriting and predicting at-risk students arouse ethical concerns. AI is capable of learning from data, leading to poor predictability and transparency.
Therefore, it is of great practical significance to prospectively study and avoid ethical dilemmas and value conflicts involving AIED. This paper tends to inquire about two research questions:
RQ1: Why are thought experiments suitable tools for investigating value conflicts arousing by applying AIED?
RQ2：How do we create and report high-quality thought experiments as tools to investigate value conflicts in the application of AIED?
Descriptive ethical theory and thought experiment methods provide a way to break through the above dilemma. Based on descriptive ethics, this study takes the value conflict as the starting point, assuming thought experiments as an effective tool to investigate value conflict. We create a thought experiment involving a humanoid teaching robot as an example to explore teachers' value orientation and potential adjustment methods in the value conflict situation to provide insights and materials for future empirical studies.
A thought experiment is a rational thinking activity carried out in the human mind following the format of the experiment, through the development of things in a hypothetical situation and the discussion of the results to find the answer to a question. Simply put, the researcher asks himself or the subject a series of "what if" questions to find the answer (Cooper, 2005).
The most crucial feature of thought experiments is that the conditions are idealized, which has many advantages over physical experiments, like avoiding the practical technical difficulties of forward-looking analysis and prediction. In this study, a thought experiment was designed to explore the four potential value conflicts of a humanoid educational robot without actually creating one to reduce or even eliminate the risk of trial and error (Norton, 1991).
The key to using the thought experiment approach is to create high-quality ethical dilemmas through fictional vignettes to explore the value conflicts and the orientations of the stakeholders in the ethical dilemmas.
A highly automated humanoid educational robot, Bo, capable of predicting high-risk students and assisting individual learning, was conceptualized for this thought experiment. It is a relatively perfect robot with high prediction accuracy and is adored by students to enhance the intensity of value conflict teachers face. Four Vignettes are created respectively based on the conflict between accuracy and transparency, educational well-being and privacy, within autonomy, and fairness.
Vignettes Background： You are the 8th-grade teacher of middle school X. This semester, the principal assigned an educational robot, Bo, to your classroom. It has an anthropomorphic appearance, collects data from the classroom through cameras, microphones and sensors, shows expressions on the screen, and communicates with you and your students through speakers. The technician who developed Bo tells you that Bo can play two roles. First, Bo will synthesize the data it collects and predict that students will likely fail the next exam. Second, Bo can tutor students on individual assignments and group discussion activities, giving tips and guidance. After half a semester of adaptation, you find that Bo’s predictions are highly accurate, and every time the students tagged by Bo are indeed the students who are struggling in the class. In addition, many students in the class overwhelmingly enjoy having Bo tutor them, finding it refreshing and fun. However, a few things have happened in the recent week to make you confused.
Vignette 1(conflict between accuracy and transparency): After school on Monday, you are writing today's classroom journal and doing tomorrow's lesson planning. Bo approaches you and says, "Mary is in a 'high risk' situation and will probably fail the next test." Nevertheless, Mary has always struck you as an obedient and conscientious student.
After listening to Bo, what would you do?
Vignette 2(conflict between educational well-being and privacy): The technician who developed the Bo is involved in the Tuesday teacher meeting. During your conversation with him, you learn that the Bo collects numerous data: there is data on students' classroom performance and grades, as well as data on students' home situations, life experiences, and many data on teachers, including the performance of your previous classes, your classroom speech, behavior, and so on. After listening to his presentation, what would be your first feeling?
Vignette 3(conflict within the autonomy): In Thursday's class, you ask students to start a discussion around rhetoric in literary work. As a teaching assistant, Bo is making the class rounds. Students in one group begin to recall the text they learned last year, and Bo says, "Your discussion seems to have strayed from today's topic." Then Bo went to another group and found that one of the students, Sam, was barely speaking, just writing something on his own, so Bo asked, "Sam, do you have any good ideas?" Sam pretended not to hear and continued to write with his head in the sand. Finally, Bo walked up to him and raised his voice to ask again, which seemed to anger Sam, who said, "I have nothing to say", then turned his head and walked to the other group. Do you think there is a problem with what Bo did when you saw the above scene?
Vignette 4(conflict within the fairness): Daisy, a student in your class, has dyslexia and often takes more time than others to organize her language. In Friday's study hall, Bo tutors Daisy on her math homework. After Daisy finished her solution, Bo showed a sad face on the screen and said, "Sorry, I did not understand. Please repeat it." The expression on Daisy's face was a bit confused. What will you do after you see this scene?

Authors

• yuan shen, Graduate School of Education, Peking University

• Qiong Wang, Graduate School of Education, Peking University