Light assignments

Assignment 1: Reflections on personal AI use

Submission due by Wed, Jan 11 with a presentation in class that day.

In this assignment, your goal is to reflect on your personal usage of AI applications / tools to share with the class. Your goal is to develop a 3-4 minute presentation for class. We will try and reserve 1 min for QA. You can approach this assignment a number of ways. Feel free to be creative! Here are some example ways of completing the assignment:

• you could take a data-driven approach to track or measure some aspect of your reliance on an AI application for a week.
• You could do a retrospective analysis of your own interactions with AI systems or that of a community that you are active in.
• You could spend a week attempting to interact with an AI model in some way, such as switching to a new technology and reporting back on the experience.
• You could interview or survey an AI engineer or AI product designer.
• You could talk about the possible societal or behavioral implications of a new emerging technology.

Assignment 2: Hack something useful for someone else

Submission due by Wed, Jan 11 with a presentation in class that day.

In this assignment, your goal is to hack together an AI application and put it in front of others. Similar to the first assignment, you will create a 3-4 minute presentation of your work. Your prototype doesn’t need to be polished or completely functional. Your aim is to identify a possible unique use case where an AI automation can augment human intelligence or behavior. There are, again, a number of ways you can tackle this project:

• You can identify a creative way to prompt GPT or ChatGPT and enable a new application.
• You can create some form of art with DALL-E and present its significance to us.
• You can prototype a pipeline to pipe online questions from different domains into LLMs and study how their responses differ from human answers to those questions posted online.
• You can finetune an existing model to support some user need.
• You can take existing models and put them in new situations to study emergent user behaviors.
• You can create a probe to study any AI product with an API.