Instructor Preview

A short guide for colleagues reviewing the MATH 346 redesign prototype.
Tip

For the current development state, decisions, and next milestone, see the instructor progress page.

Why This Prototype Exists

I am preparing to teach MATH 346, and I do not think that teaching the course exactly as it was taught before is the right approach in the LLM era.

Students will use AI tools. The course should therefore teach them how to program, test, debug, interpret results, and critically inspect AI-generated code or explanations. The goal is not to make the course about AI. The goal is to teach mathematical and statistical software in a world where AI assistance is part of the working environment.

How The Prototype Was Built

I used the existing course slides as a starting point for the syllabus topics and progression. With Codex, I then prepared this prototype website, rewriting the materials around problem-focused computational work, validation habits, and responsible AI use.

The site does not redistribute copyrighted slide decks, textbook PDFs, answer keys, exams, or private student work. It contains original course materials and starter code prepared for this redesign.

What To Review First

Weeks 1-4 are the most developed parts of the prototype. They show the intended teaching pattern:

Week Focus Suggested review
Week 1 MATLAB arrays, scripts, and plots Look at the survival checklist, lab, and AI critique activity.
Week 2 Functions, conditions, and tests Look at the testing emphasis and AI debugging activity.
Week 3 Curve fitting, interpolation, and residual checks Look at the model validation and overfitting critique.
Week 4 Numerical methods and 3D plotting Look at the numerical checks, surface plots, and AI answer review.

Weeks 1-4 have now been validated as the prototype pattern for the course. Weeks 5-15 are still more schematic. They show the intended arc of the course, but they are not yet expanded to the same level as Weeks 1-4.

What Feedback Would Help

I would especially appreciate feedback on these questions:

  • Does the redesign still respect the spirit and content of the existing course?
  • Are Weeks 1-4 realistic for two 75-minute meetings per week?
  • Is the balance right between MATLAB/R skills, mathematical thinking, and validation?
  • Do the LLM-related activities feel pedagogically useful rather than distracting?
  • Are there essential commands, concepts, habits, or warnings missing from the first four weeks?
  • What should be adjusted before expanding Weeks 5-15 in the same style?

The Intended Shift

The old version of the course naturally emphasized learning MATLAB and R command by command. That made sense before LLMs were widely available.

In this version, students still learn MATLAB and R, but each week is organized around a computational question:

  1. What is the problem?
  2. What representation or model should we use?
  3. What code answers the question?
  4. How do we know the result is correct?
  5. What can an LLM help with, and what must we still verify ourselves?

The core skill I want students to leave with is not just the ability to run software. It is the ability to produce a computational result that they can defend.