Students will critique AI tools’ output to create representative images of the relevant cultures by applying prompt engineering.

Theme: Assessment, Research, AI Literacy

Tools: Poe AI, ChatGPT

Recommended Disciplines: Communication, Linguistics, Cultural Anthropology, Sociology, Political Science, and Women and Gender Studies

Context: An undergraduate class to provide practical tools to manage intercultural encounters in the context of the US

Recommended Course Size: 20-26 students

Roles: Teacher as a facilitator, Students as researchers

Classroom Time Needed: 25 minutes + 10 minutes for presentation and reflection

LEARNING OBJECTIVES:

Practice prompt engineering to evaluate the accuracy and appropriacy of the visual AI outputs

PRE-WORK

  1. Before classroom time, ask students to familiarize themselves with the AI tools to be used. (To hold them accountable, start the class with a lead-in activity. E.g. a Mentimeter activity asking them to share what they discovered while they are playing with the tools.)
  2. Provide a brief definition of AI and prompt engineering, and how generative AI and prompts work either as a syllabus statement or verbally to facilitate students’ use of the tool and create awareness about the importance of prompting through hands-on practices.

“AI is a simulation of human intelligence that works using algorithms and huge amounts of data. This dataset emerges through the accumulation of online human-produced data and algorithms, which are step-by-step mathematical calculations creating connections among different data (Striphas, 2015) to generate the searched information. Prompting can be likened to whispering into Alaadin’s magic lamp, in other words; it is the process of “designing, crafting, and refining” the prompts you share with AI to elicit the appropriate outputs (p. 2, Bozkurt & Sharma, 2023).

 

Considering the huge amount of data, we can most probably find the answers to our questions. However, it is important to know how to use AI to find these answers. At this point, prompting is the key. While prompting we should keep in mind that even small details such as the order, capitalization, and spacing can affect the output (Butler et al. 2023). Therefore, prompting may pose some challenges. To overcome these challenges, you can:

  • Define your objective: What is the output you are aiming to have?
  • Familiarize yourself with the tool: What are the pros and cons of using that tool? What type of prompts yields the best outputs from this tool?
  • Be precise: write straightforward instructions and avoid complex and ambiguous language.
  • Provide context: answer who/what/how/when/why questions in your prompt.
  • Refine and iterate the prompt: if your prompt does not yield the desired output, you can refine your prompt by using different wording, word order, spacing, and capitalization. Or, you can also iterate the same instruction because you should keep in mind that the same prompt can generate different outputs while using AI.
  • Specify the format: Are you looking for an image/text/video? Do you want the output in a paragraph format or a list? Do you want single results or multiple results?
  • Include key details: What are the essential features you are looking for in the desired output? (e.g. statistics, people, resources, numbers, etc.)
  • Test and iterate: apply different prompts and evaluate the outputs. Iterate this process and refine your input based on the AI-generated responses you receive.
  • Consider ethics: Critique the output. Is it biased? Does it reproduce racist/sexist/politically biased convictions? Does it create discriminatory discourses? After evaluating the AI-generated output, you can decide if you want to use the output or not (Bozkurt & Sharma, 2023).

 

INSTRUCTIONS

  1. Depending on the classroom dynamic; divide the classroom into two, or ask students to self-select in two groups.
  • Group A: uses Poe and/or ChapGPT.
  • Group B: does not use any AI tool.
  1. Ask students to work in their groups and create a gallery of images that represent the American Culture.
  2. At the end of 25 minutes, ask students to present their gallery to the whole class by referring to these questions:
  • Which online tools did you use?
  • What are your research prompts?
  • Why did you add this visual? In what ways do you think this represents/ does not represent American culture?
  1. Following the presentations, ask students to reflect on their process of image research with/without using AI (OpenAI, 2024):
  • How user-friendly did you find the tools you use for researching images?
  • What specific features or aspects made the process easier or more challenging?
  • How would you rate the efficiency of the tools you used?
  • Do you think that there is a significant difference in the quality and relevance of the images in the two galleries?
  • Do you think that the images in the two galleries represent the American culture or do they iterate the existing and non-reflective biases?

 

REFERENCES

 

Bozkurt, A., & Sharma, R. C. (2023). Generative AI and prompt engineering: The art of whispering to let the genie out of the algorithmic world. Asian Journal of Distance Education, 18(2), 1-7.

 

Butler, J., Jaffe, S., Baym, N., Czerwinski, M., Iqbal, S., Nowak, K., Rintel, R., Sellen, A., Vorvoreanu, M., Hecht, B., & Teevan, J. (2023). Microsoft New Future of Work Report 2023. Microsoft Research Tech Report MSR-TR-2023-34 (https://aka.ms/nfw2023).

 

OpenAI. (2024). Reflection questions on the image research process with/without using AI. Retrieved from https://chat.openai.com/

 

Striphas, T. (2015). Algorithmic culture. European Journal of Cultural Studies, 18(4-5), 395-412.

 

 

List Author(s) Name Here (required): Mervenur Cetin
Photo Source (required): John Schnobrich, https://unsplash.com/photos/three-person-pointing-the-silver-laptop-computer-2FPjlAyMQTA
Image Alternate Text: A group of students working together

Edit Link: (emailed to author)
Request Now