Posit Academy in the Age of Generative AI

Lessons from the Frontlines

James Wade

Or… How I learned I was wrong about GenAI

Posit Academy at Dow

• 200 researchers at Dow have gone through Academy
• Chemists, chemical engineers, physicists, and more
• Mostly even split between R and python
• Growing community of citizen data scientists

GenAI as Rocket Fuel 🚀

{chattr}

{gptstudio}

Github Copilot

Over half of Posit Academy learners gave AI code assistants a 2 star rating or less.

What were the most rewarding high-growth periods of your life?

1. Learning wasn’t the point.
2. Learning really worked.

Andy Matuschak | How might we learn?

So what’s missing?

Threshold Concepts

What Are Threshold Concepts?

These are ideas that, once understood, transform your perception and approach of a discipline.

Threshold concepts must be encountered.

Threshold concepts: embrace the struggle

• Troublesome: Conceptually difficult, challenge prior knowledge
• Liminal: Understanding is non-linear and maybe confusing
• Transformative: Recognize new patterns, a new understanding
• Irreversible: Once understood, you cannot unlearn it

My rewarding, high-growth period

1. Asking for help means you care
2. Self-care is not selfish
3. Incomplete todo lists are okay

Threshold Concepts in Data Science

1. Tidy data enables efficient analysis
2. Modular code enhances reusability & clarity
3. Visualization is a tool for exploration & communication

What ChatGPT won’t give you

Learning dispositions for code-based data science

1. Sense of the possible

   An internal sense of capability and persistence

2. Flexibility

   Expand your palette of possibilities

Time to discard AI coding assistants?

The famous “bicycle for the mind” metaphor is better because it has no agenda other than the one you bring. It just lets you reach a wider range of destinations than you could on foot. And it makes the journey fun too, particularly if you’re biking along with some friends. The bicycle asks: where do you want to go?

Of course, that question assumes your destination is well-known and clearly charted on some map. But those most rewarding high-growth experiences are often centered on a creative project. You’re trying to get somewhere no one’s ever gone before—to reach the frontier, then start charting links into the unknown. Learning in service of creation. It’s a dynamic, context-laden kind of learning. It’s about more than just efficiency and correctness. More than just faster gears on a bike. That’s the kind of learning I feel an almost moral imperative to help create. Source: https://andymatuschak.org/hmwl/

Coding as a

contingent discipline

• We discover what we want to build as we build it –> so we should use ChatGPT to shorten the iteration cycles

Posit Academy Foundations Curriculum

R Basics

Read & Visualize

Summarize Data

Create & Select Columns

Tidy & Join Tables

Write Functions

Work with Data Types

Analyze Data

Showcase Results

Iterate

How to incorporate AI code assitants?

1

2

3

4

5

6

7

8

9

10

How to incorporate AI code assitants?

Week 3: Summarize Data

1

2

3

4

5

6

7

8

9

10

“Explain this code piece by piece”
“Show me how to group by two columns”

How to incorporate AI code assitants?

Week 6: Write Functions

1

2

3

4

5

6

7

8

9

10

“Add a roxygen skeleton to my code” “Suggest 3 ways to improve my code”

How to incorporate AI code assitants?

Week 9: Showcase Results

1

2

3

4

5

6

7

8

9

10

Try out code assistants in the IDE “What pkgs should I explore next?”

What are TCs for coding assistants?

Some ideas…

1. Drive faster but don’t forget to steer
2. Prompting matters: learning to use these tools is a skill
3. …?

---

Are these right? What’s missing?

🙏 Thank you to those who make it happen! 🙏

• Tony Sokolov
• Michael Hausinger
• Lynette Naler
• Dow (Apprentice) Mentors
• Dow’s Academy Learners
• Dow Leaders

• Lindsay Lintelman
• Kristin Bott
• Ibi Togun
• Brian Law
• Eric Pite
• Many Academy Mentors

Slides: https://jameshwade.github.io/posit-conf-2024