Jill Purdy
Classroom Policies
These policies appear together on the syllabus:
Academic Standards: Students are expected to meet the traditional standards of honesty and truthfulness in all aspects of their academic work at UW Tacoma. All work submitted to an instructor in fulfillment of course assignments, including papers and projects, written and oral examinations, and oral presentations and reports, must be free of plagiarism. Plagiarism is using the creations, ideas, or words of someone else without formally acknowledging the author or source through proper use of quotation marks, references and the like. Student work in which plagiarism occurs will not be accepted as satisfactory by the instructor and may lead to disciplinary action against the student submitting it. Any student who is uncertain whether their use of the work of others constitutes plagiarism should consult the course instructor for guidance before formally submitting the work involved. Please review the academic honesty page at the link above.
AI/Machine Learning Tools: For this course, ChatGPT and similar Artificial Intelligence tools are considered sources of information. As such, those sources must be cited if they are used in the creation of your work, making it clear what has been originally contributed by you and what was taken from an AI tool. As noted above, all submitted work is expected to be original, and your research and analysis should go well beyond what AI tools produce. Other courses may have different policies based on different learning objectives, so be sure to understand the expectations for each class.
Assignment/Activity
Class discussion
What can Large Language Model (LLM) AI do for us? What can it not do?
(This discussion focuses on large language models such as Chat GPT and Claude rather than other forms of AI.)
Learning Outcomes
Describe the possibilities and limitations of LLM AI
Identify benefits and costs for individuals of using LLM AI
Identify benefits and costs for society of using AI
Tools/Resources Used
ChatGPT LLM: https://chat.openai.com
Claude LLM: https://claude.ai/
Diversity in AI: https://aiindex.stanford.edu/wp-content/uploads/2021/03/2021-AI-Index-Report-_Chapter-6.pdf
Approximate Time to Complete
20-30 minutes
Step-by-step Instructions
The instructor begins by sharing an example of successfully using AI to be more productive, such as using Chat GPT to create a first draft of a project description. The instructor notes several benefits: (1) creating a prompt for AI to produce text required thinking through the project carefully, (2) reading the first response from AI prompted further thinking and description to include elements that were left out, and (3) it was easier to edit a first draft created by AI than it was to write all the words initially. Quite a bit of editing was needed to make the draft good according to the instructor’s standards. The final document was about 75% instructor’s text and about 25% AI-generated text.
Students are asked to share ways in which they have used AI and what challenges, if any, they have encountered.
Instructor shares examples where AI has given poor results or inaccurate information, e.g. making up legal cases that do not exist. Explain that models may use a predictive text approach that may replicate sequences of words regardless of their accuracy. Statistics and data must always be checked. What comes out of AI models is limited by what they were trained on, and training/inputs may be limited or may stop at a given time.
Questions for discussion:
How is AI being used in your classes? Why do you think some teachers don’t want you to use AI for doing homework or writing papers?
If you wrote and copyrighted a book, or music, or piece of art, would you want AI to be trained on it?
A significant majority of AI programmers are male in the United States and about two-thirds are white. Do you think this has any effect on how AI operates?
Microsoft invested $10 billion in ChatGPT expecting that it will make them money. Many AI models charge fees to use them. What do you think the future holds for equitable access to AI?
Conclusion: Many different types of AI are being used in the world today and a growing number of jobs involve using AI to assist with tasks, increase productivity, and support innovation. When new technology is introduced, the framework of legal and ethical protections usually lags behind by years. We are in the time right now when we as a society need to think about how we want machine learning to behave.
Reflections on Creating the Assignment
Student reactions were widely varied, with some having little to no familiarity with LLM AI tools, and others who have begun using them as employees.
Post-Implementation/Testing Reflection
What about your project worked well? What would you revise for future iterations?
Ask students to complete a task using an LLM AI tool so that all students have a baseline familiarity with prompts and outputs.
When you tested this policy or assignment, how did you feel working through the steps?
This was an energizing topic that generated good discussion. Some students seem a bit wary talking about it freely with a professor present given the different attitudes among faculty towards AI.