trustworthyAI

Assignment 4: Using AI in Ways That Matter to You

Due: Monday, April 6, 2026, Noon, Eastern time.

This is an individual project.

Objective

This assignment provides an opportunity to explore a selected dimension of AI to try to do something (instead of just highlight places where things get broken). I’ve given five options designed to investigate AI capabilities, trustworthiness, ethical implications, or creative potential in distinct contexts; you are 100% welcome to go in a different direction of similar scope.

Instructions

Select one of the following options:

1. AI as a Collaborative Creative Partner

Objective: Use AI for your creativity in storytelling, art, or design, and evaluate your feeling about this creative collaboration and trust.

• Select a creative process (storytelling, visual art, game design, music composition, etc.).
• Design prompts for GPT to generate creative content collaboratively with you.
• Document how GPT influenced your creative decisions or enhanced your process.
• Discuss your interaction with the AI.

2. Analyzing GPT’s Understanding of Humor and Sarcasm

Objective: Evaluate GPT’s ability to understand nuanced language such as humor, irony, or sarcasm.

• Craft prompts that explicitly involve humor, sarcasm, or irony, varying their subtlety and context (context can include technical domains, other languages or dialects, etc.).
• Collect and analyze GPT responses, highlighting examples of correct or incorrect understanding.
• Discuss implications of these misunderstandings in specific sensitive or high-impact situations (e.g., mental health, customer service, social interactions).

3. AI in Scientific Research

Objective: Explore how GPT-like models support your scientific tasks such as hypothesis generation, literature reviews, or experimental design, examining trustworthiness, accuracy, and limitations.

• In research that you care about and work on, integrate AI into your processes sometime in a coherent way in the next week.
• Record how you use GPT to assist with generating hypotheses, summarizing literature, or proposing experimental designs.
• Critically evaluate GPT’s output, discussing accuracy, reliability, and potential pitfalls or biases.

4. Evaluating GPT’s Common Sense

Objective: Critically evaluate GPT’s robustness in applying common-sense reasoning to everyday scenarios (perhaps, if you’re willing, things in your real daily life) and analyze implications for trustworthy AI interactions.

• Construct prompts that require common-sense reasoning, intentionally including ambiguous or nuanced situations.
• Document GPT responses, highlighting correct and incorrect reasoning.
• Discuss specific reasoning failures and reflect on their broader implications for deploying AI responsibly in real-world applications.

5. Structured Literature Review (mandated for DTAIS Fellow)

https://github.com/pless/trustworthyAI/blob/main/Project4_LiteratureReview.md

Write-Up

You will turn in a blog post approximately “3 screenfuls” in length, discussing:

1. Chosen theme and prompt: Explain your selected option, detailing your chosen task and your designed prompts.
2. Presentation of AI responses: Showcase key examples of GPT’s output, highlighting strengths, weaknesses, and variability.
3. Analysis: Critically analyze GPT’s performance, clearly discussing trustworthiness, accuracy, implications, or creative influence, based on your chosen theme.
4. Conclusions: Share insights gained, including any shifts in your perspective on GPT’s strengths, weaknesses, ethical considerations, or role in trustworthy AI applications.

You may work in groups of up to 3 students. One group member must publish your write-up as a blog post. You may use GWU’s blog service (https://blogs.gwu.edu/) or platforms such as GitHub, WordPress, etc.

Submit your blog or PDF through the following form:

Assignment Submission Form