Getting Started with AI Coding and Notebooks

Date: June 17, 2026

This is part of the MS-CC AI Readiness Webinar Series

Researchers and practitioners increasingly use tools like ChatGPT and Claude to explore datasets and answer questions. But what happens when you need to:

• Analyze hundreds or thousands of records

• Repeat analyses on new data

• Share methods with collaborators

• Document exactly how results were generated

• Create workflows that others can reproduce

This webinar explores when conversational AI is sufficient and when notebooks become a more powerful and sustainable approach.

---

The Guiding Question

If I can upload a CSV to ChatGPT and get an answer, why would I use a notebook?

The answer is not that notebooks are always better.

Different approaches are appropriate for different tasks. This webinar introduces a framework for deciding:

• When a conversational AI interface is enough

• When notebooks add value

• When APIs enable automation at scale

• When open-source models may be useful

---

A Real-World Scenario

Imagine your institution conducts an AI readiness survey and collects responses from faculty, staff, researchers, and students.

Your goal is not simply to analyze the dataset.

Your goal is to answer questions such as:

• What AI capabilities are people already using?

• What concerns do different groups have?

• What types of support are needed?

• How should institutional resources and training be prioritized?

Throughout the webinar, we will use this scenario to demonstrate multiple approaches for moving from:

Data → Questions → Answers → Action

---

What You'll Learn

Level 1: Conversational AI

Use ChatGPT or Claude for:

• Rapid exploration

• Asking questions about a dataset

• Prototyping ideas

Best for: One-time analyses and exploratory work.

---

Level 2: Reproducible Notebooks

Use notebooks to:

• Document your workflow

• Create visualizations

• Repeat analyses on new data

• Share methods with collaborators

Best for: Research workflows that need transparency and reproducibility.

---

Level 3: AI APIs

Use notebooks with APIs to:

• Process hundreds or thousands of records

• Apply the same analytical logic consistently

• Automate repetitive analysis tasks

Best for: Scaling AI beyond a single conversation.

---

Level 4: Open-Source Models

Use local models when you need:

• Greater control

• Data privacy

• Customization

• Reduced dependence on commercial APIs

Best for: Specialized research workflows and sensitive data environments.

---

Four Ways to Analyze Data

| Approach | Ease of Use | Scalable | Reproducible Workflow | Best Use Case |

|-----------|-------------|-----------|----------------------|----------------|

| ChatGPT / Claude | High | Limited | Limited | Exploration and brainstorming |

| Notebook (No AI) | Moderate | Moderate | High | Transparent, documented analysis |

| Notebook + AI API | Moderate | High | High | Automated analysis workflows |

| Notebook + Open-Source Models | Moderate | High | High | Privacy, control, customization |

---

The Real Value of Notebooks

The question is not:

Can AI answer my question?

The question is:

Can I reproduce, share, automate, and extend this analysis six months from now?

A notebook turns analysis into a documented workflow:

load_data()

analyze()

visualize()

report()

The workflow can be rerun, modified, shared, and reused.

That is why notebooks remain valuable in the age of generative AI.

---

What You Should Leave Understanding

By the end of this session, you should be able to answer:

• When is ChatGPT enough?

• When do notebooks become useful?

• When should I automate analysis with APIs?

• When might an open-source model make sense?

The goal is not to turn everyone into machine learning engineers.

The goal is to help you become an informed AI practitioner who can choose the right tool for the right problem.