AI Integration (Front & Center)

AI tools such as ChatGPT are an integral part of this course. You are expected to leverage them thoughtfully across paper discussions, coding labs, and the final project—while taking responsibility for verification and integrity.

How AI is used in this course

• Paper discussions: Presenters and discussion starters must submit an AI_TRACE.md capturing how AI assisted (summaries, question generation, code snippets to illustrate equations, etc.). In-class discussion may produce additional AI_TRACE.md notes—upload those to GitHub as well.
• Coding labs: Each team includes a GPT Usage Log within their lab report documenting and reflecting on AI-in-the-research-loop:
  • What prompts did you give ChatGPT?
  • Which outputs/code did you accept, and how did you validate it?
  • Any pitfalls?
  • What are good use cases for LLMs and what fails consistently?
• Projects: Proposal and final report include an AI assistance section (or AI_TRACE.md) detailing prompts, accepted suggestions, and verification steps.

Expectations

• Document all AI assistance (prompts, outputs, rationale) in AI_TRACE.md or the specified GPT Usage Log sections.
• Verify all AI-assisted content (derivations, code, claims) with tests, references, and physics checks.
• Be critical: discuss where AI helped vs. misled you.
• Evaluation: reflection logs will be assessed for their thoughtfulness and are expected to demonstrate iterative, critical, and discerning use of AI — not simple “solve this for me” requests.

Coursework & Grading

Grade Breakdown

• Paper Presentations — 20%
• In-Class Participation & Reflection Logs — 10%
• Coding Labs — 30% (4 labs; each student develops in 2 labs which amount to 20% & reviews in 2 which amount to 10%)
• Project Proposal (1-2 pages) — 10%
• Final Project Report (3-5 pages) — 30%

Participation (weekly)

• Prep: Submit a short GPT-assisted reflection or 1-page critique + 1–2 discussion questions in the repo by Monday 23:59.
• In-class: Quality of contributions, discussion engagement, and responses to peers.

Labs & GitHub Workflow

Team Tasks per Lab:

• T1 — Toy Scenarios: implement simple, analytically tractable cases. How do these toy scenarios help us understand the mechanics?
• T2 — Realistic Scenario: complex/realistic setting; compare with standard or state-of-the-art results; reflect on result metric and the advantages/disadvantages
• T3 — Novel Directions: explore creative extensions (e.g. new constraints or regimes) that the authors did not explore.

Submission window: Developer PRs due by Monday 23:59 before the Tuesday interactive session. Reviewers clone the repo and run tests during the interactive session. After the session, reviewers submit reviews/followup.md by end of week.

GitHub: How We Submit

1. Branch: create labXX branch; commit code, plots, tests, templates.
2. PR: open [LABXX] Team T# — names using the PR template. PR is the formal submission for review & grading.
3. Review: assigned peers use the issue template to review; developers respond and iterate.
4. Tag: after addressing reviews, tag labXX-submit.

Paper Week AI_TRACE.md Requirements

• Presenters: upload AI_TRACE.md outlining AI assistance (summaries, math/code drafting, figures).
• Discussion starters/respondents: upload AI_TRACE.md with prompts and how they shaped your questions/talking points.
• In-class additions: any AI-assisted prompts or outputs arising during discussion should be committed as an additional AI_TRACE.md or appended to the existing file.

Schedule (Fall 2025)

Week	Date	Topic / Format	Notes	Submission
1	Tue 02/09	Course overview	GitHub + AI_TRACE + PR workflow	—
2	Tue 09/09	Physics-Informed Neural Networks (PINNs)	Paper discussion	—
3	Tue 16/09	When & Why PINNs Fail to Train	Paper discussion	—
4	Tue 23/09	Lab 1 Interactive Session — PINNs	Hands-on workshop	Lab 1 code due Mon 22/09, 23:59
5	Tue 30/09	Neural Radiance Fields (NeRF) / Fourier Features	Paper discussion	—
6	Tue 07/10	CryoDRGN: Reconstruction from Cryo-EM	Paper discussion	—
7	Tue 14/10	Lab 2 Interactive Session — NeRF	Hands-on workshop	Lab 2 code due Mon 13/10, 23:59
8	Tue 21/10	Differential Walk on Spheres	Paper discussion	—
—	Tue 28/10	Reading Week	No class	—
9	Tue 04/11	Lab 3 Interactive Session — MCMC Rendering / Walk on Spheres	Hands-on workshop	Lab 3 code; Project Proposals due Tue 04/11, 23:59
10	Tue 11/11	Fourier Neural Operators (FNO) / Deep Operator Networks (DeepONet)	Paper discussion	—
11	Tue 18/11	Neural Implicit Flow	Paper discussion	—
12	Tue 25/11	Lab 4 Interactive Session — Neural Operators	Hands-on workshop	Lab 4 code due Mon 24/11, 23:59
—	Tue 02/12	End of Term		Project Report due Tue 02/12, 23:59

Policies

Critical Use of ChatGPT and AI Tools

AI tools are a core component of this course. You are encouraged to use them for brainstorming, summarizing, debugging, experiment design, and exploring alternatives. However, the focus is on critical and reflective use. You are required to document all AI assistance (prompts, accepted outputs, and validation steps) in an AI_TRACE.md or the designated GPT Usage Log sections.

• Thoughtful Integration: Use AI to assist with idea generation, code drafting, explanation, visualization, and literature exploration — but always validate outputs and document your reasoning.
• Critical Verification: Check AI-generated content against original references, perform physics-based sanity checks, and test outputs before including them in your work.
• Transparency: Clearly log how AI was used, which outputs were adopted, and how they were verified. Undocumented AI contributions are considered incomplete work.

You are fully responsible for the accuracy, originality, and integrity of your submissions. AI is a collaborator, not a substitute for your own critical thinking.

Communication & Announcements

Discord (invite on Quercus) is the main channel for coordination and informal discussions. Announcements, materials, and grading details are posted on Quercus.

Late policy

Discussion logs and Developer PRs are due by Monday 23:59 before each discussion or interactive lab. There will be a 30% deduction if you submit late, but before the start of that week's lecture (i.e., if you submit anytime between 12:00am and 4pm on Tuesday). No homework will be accepted after the start of lecture.