Production 3: Critical Algorithmic Literacies and Generative AI
Based on the readings and class discussion, utilize ChatGPT and/or other AI ‘art’/image making software to explore 1) the possibilities of meaningful learning in your teachable or discipline while 2) critically engaging the tool with regards to figuring out any algorithmic bias, ChatGPT rhetorical techniques, and programming suppositions (e.g., discern what kinds of ‘data sets’ or programmer assumptions and values are informing the interactive experience; reflect on how the tool addresses you or exerts rhetorical ‘authority’).
Here are some examples. Feel free to use or modify these examples – or come up with your own – that address points 1 and 2 above.
Example: Role Play or Engage in a Debate: If your research area, discipline/teachable is history, political science, social studies, or philosophy, prompt ChatGPT to simulate (role play) a person or historical actor from the past and engage in dialogue about a specific historical event/situation or controversy. Prompt ChatGPT to take the role in the dialogue or debate as the historical actor you select (vis a vis an event, setting, conflict, topic, etc). The identity of AI in the the role-playing ‘dialogue’ simulation can be a well-known historical figure, or you can prompt the AI to be an ordinary person who is historically situated in the event, situation, conflict, etc.
Considering trying to prompt the software into a debate over a controversy or historical issue that you know something about, so you can counter/challenge the tool with arguments or counter-arguments and see if and where the system ‘goes off the rails’.
Example: If your teachable is language learning (FSL, ELL, L2), explore if and how it is possible to utilize ChatGTP as a real-time language tutor and conversational agent. Ensure you are clear in your prompts to get ChatGPT to do any corrective or explanatory functions associated with learning a language (grammar, word choice, proper idioms). Consider prompting ChatGPT to have a conversation about a subject that is meaningful to you, where you might learn new vocabularies or idioms associated with the subject of interest.
Example: If your teachable/discipline is computer science or maths or technological education, try to get ChatGPT to help you understand or elucidate some aspect of coding or solving authentic computational problems, or helping you do a complex process in relation to your practice. Can ChatGPT be used to ‘show’ processes and provide rationale that help you learn? Is the code or process correct (can you test it)? Does the AI model and represent procedures in a clear way, or address questions you have about the process? Is anything essential left out? I’m not a math teacher, so consider searching web for other possible meaningful uses.
Example: If your teachable/discipline is English/Literature, co-author a story with ChatGPT (take turns, and in your prompt, and specify writing ‘turn’ length – e.g., ‘no more than 3 sentences’ per turn). Specify story parameters like genre and style – and consider asking ChatGPT to model specific techniques associated with the practice of writing: word choice, imagery, use of metaphor or other literary techniques, etc.
Example: If your teachable is in the arts (visual, music, etc,) consider combining aspects of example 1 (dialoging with an artist or historical figure) while analyzing generative AI art tools to explore the encoded genre conventions and assumptions in AI-generated art/images/cultural texts.
For all teachables/disciplines, consider going to an AI-art generating tool (listed below) to see how it represents any related facet from above, e.g., how does the AI-art represent the historical conflict, setting, individual, story setting, STEM process, etc, with a keen eye to any kind of algorithmic bias, ideological bias, or (mis)representation.
Expectations:
- Teachables/Disciplinary Expertise: For this production, you need to KNOW something – (have some expertise) in order to critically evaluate the AI. Test what can be done and try to ‘push’ at the limits of what the AI can do!
- Prompt Literacy: Specific, detailed prompts and thoughtful interactions are key to creatively working with AI. In your analysis please ensure you communicate what prompts you used, and if you needed to refine prompts or start over. Take screen shots of key learning moments!
- What and how did you learn through the interactive process? Did the AI hallucinate or provide (mis)information? Evaluate the learning experience and what you did in relation to the AI. Did you detect algorithmic bias? Could you make inferences about the assumptions of the programmers or training sets used? Implications for learning? Implications for academic integrity? Recommendations, risks, or warnings?
- Your analysis (essay) should include documentation (screen shots, video clips, etc) of the process; ensure that you explain what is happening in any image or screen shot you use to document process or a significant moment in your interactions with AI. Spend enough time with the AI to evaluate!
- Evaluative/Descriptive Essays should be at least 500 words – but may be longer if you need more text to communicate the experience. Make sure you address 1) significant learning potential (given proper use) 2) critical reflection on how the AI ‘works’ (rhetoric, bias, what’s included/excluded, Eliza effect, language norms, how you are ‘addressed’ by the AI, etc — see the slide deck and article if you need a refresher).
Examples of conversational/generative and multimedia generative AI software include:
ChatGPT (OpenAI)
DALL-E 2 (OpenAI)
Others: Note I have not tried all of these and some may not be useful or accessible.
Bing Image Creator (Microsoft), Canva AI Image Generator, CM3leon (Meta), Imagen AI (Google), Midjourney, Modyfi, OpenArt, Stable Diffusion (and its new version, SDXL 1.0)
Music (creating/editing): MusicLM (Google), MusicGen (Meta), Musicfy, Stable Audio
PLEASE IGNORE ANY LINKS TO OTHER PRODUCTIONS BELOW