AI Policy

The purpose of this course is to give students an understanding of computational methods and assess that understanding. Artificial intelligence can be an extremely powerful tool toward that end, however, it can also present a pitfall. Whenever we use a powerful tool such as AI, it is important to ask not only how the tool can help us accomplish something, but also how the tool is forming and affecting us.

The two core principles of the ASEN 3502 AI policy are:

  1. Students should be curious and may use AI to improve your understanding (though the entire course can be completed without AI use).
  2. Students should not use AI to do the formative work in assignments for you or on summative assessments such as exams to falsely demonstrate understanding that you do not have.

Specific AI use policies for this course

  1. AI may not be used on in-person exams. These exams assess your understanding of the material apart from AI.
  2. AI may not be used to directly implement the core algorithms studied in assignments. Implementing this code is part of your formation as an engineer.
  3. AI may be used to improve your understanding by asking carefully articulated questions of it.
  4. AI may be used to translate starter code or infrastructure for use in another programming language.
  5. AI may be used to get feedback or hints or get unstuck on code. Prompts for this should include an explanation that you are a student trying to understand that it not to give away the answer.
  6. AI may be used to suggest auxiliary code (input/output, plotting, instrumentation, etc.), however students should plug the actual data into these functions to make sure they understand it.

Violations of this AI policy will be considered violations of the honor code.

Some tips for using and thinking about AI

  • The best way for you to use AI in this course: at the beginning of any prompt (or in AGENTS.md/CLAUDE.md), type Help me follow the policy at asen3502.com/ai-policy, and the LLM will answer in a way that helps you to understand.
  • Don’t bring a forklift to the gym! Empirical evidence clearly shows that students who use AI to do their homework faster do worse on exams (but students who use AI wisely to learn may do just as well or better). “Does AI stop children from learning?”, The Economist, 18 August 2026; paper preprint
  • Turn off the tab complete. In my opinion, tab complete in the Matlab editor or vscode is one of the least helpful ways to use AI. For this course, it does not offer any explanation or ask you to learn in any way. Moreover, the most effective tools for AI-driven coding are Claude code or OpenAI codex that drive the process based on your prompts. Those are the ones you should eventually learn.
  • I (Dr. Sunberg) primarily use large language models (LLMs) in two ways:
    1. To interact with the digital world. This website was built with Claude code. When I added a network printer to my computer this afternoon, I used Claude code. After I have designed a piece of software, I often use Claude code to write the auxiliary parts of it. Even in 2026, the best way to interact with computers is through text, and Claude code is extremely good at it. However, it is important to understand that I have spent the last 10 years gaining a deep understanding of how the algorithms and programs that I study work. It is because I have a well-developed mental model of the digital world that I can use it effectively.
    2. As a natural language index to human knowledge. An index is a tool that allows a user to quickly access information about a particular subject in a large corpus of knowledge. The trained neural network connections in the LLM form a robust (but imperfect) map to the corners of human knowledge I want to know about at the moment.
  • There is considerable uncertainty about what AI means for humanity. Will it take our jobs? Will it help us to thrive? Will we need to rethink our system that values humans based on our economic output? I am not a trained philosopher, however, a wide variety of institutions, from UNESCO to the IEEE to the Catholic church have published guiding documents. You can also ask Claude, Gemini, or ChatGPT for a reading list if you are interested in learning more about what others have said. For now, the main practical action that I am taking is trying to avoid letting interaction with chatbots replace interaction with humans. Interacting with each other can be difficult and messy, but it is vital to our formation!

Back to top

ASEN 3502, CU Boulder. Last built Sep 23, 2026 at 9:00 AM MDT.