Using AI Tools in Data Science 🤖

---

layout: true
  
<div class="my-footer">
<span>
<a href="https://DataScience4Psych.github.io/DataScience4Psych/" target="_blank">Data Science for Psychologists</a>
</span>
</div>

---

# When and how to use AI in this course

---

## Learning R is challenging

.pull-left[
.hand[Common frustrations:]
- Syntax feels foreign
- Error messages are cryptic
- Many approaches to same problem
- Documentation can be confusing
- I already know how to do this in SPSS
- R is a whiny baby
]

---

## Why not use AI from the start?

.question.med[
If AI can write code for us, why learn `\(R\)` the hard way first?
]

.hand[Because AI tools are not perfect]:
- Can give incorrect code
- May use outdated approaches
- Don't always follow best practices
- Require R knowledge to evaluate
- Can't teach you how to think like a programmer
- You can't argue with them if you don't know enough to get annoyed by their antics

---

## Productivity vs Learning

.pull-left[
What research shows:
- AI can boost productivity at all levels in writing (Noy et al, 2023) and coding (Chen et al., 2023; Peng et al 2023)
- These gains are seen in both novices and experts
- The short-term productivity gains can be substantial for novices
- Limited research on how AI can impair the transition from novice to expert 
- Open question: How can AI be used to help novices become experts?
]

---

## Four stages of AI use

- My advice for how to learn R with AI (Garrison & Tyson, 2026)
<ol start="0">
<li>No AI - Building foundations</li>
<li>AI for understanding code</li>
<li>AI for debugging help</li>
<li>AI for code generation</li>
</ol>

.footnote[Garrison, S. M. & Tyson, H. K. (2026). *Pragmatic AI Integration in R Teaching: From Frustration to Facilitation* In I. Katzarska-Miller, M. Jackson, & M. Fortner, (Eds.) Integrating Artificial Intelligence into the Psychology Classroom. Society for the Teaching of Psychology: teach-
psych.org/ebooks/]

--
  
.question[
    How will you know when you're ready for each stage?
]

---

## Stage 0: Building foundations
.pull-left[
What you'll learn first:
- Basic R syntax
- Data structures
- Functions
- Debugging strategies
]
--
  
.pull-left[
```r
# You'll write code like this
    diamonds %>%
      select(carat, price) %>%
      arrange(desc(price))
```
]

.pull-right[
And understand:
- What each line does
- Why it's structured this way
- How to fix errors
]

---

## Stage 1: AI for understanding

.pull-left[
Once you can write basic code, you can use AI to:
- Get line-by-line explanations
- Understand error messages
- Learn alternative approaches
]

--
.pull-right[
But first verify:
- Does the explanation match your understanding?
- Are there terms you should look up?
- Does the suggested usage align with class materials?
]

--
.pull-left-wide[

``` r
# Ask AI: "What does this code do?"
ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point() +
  geom_smooth()
```
]

.center.pull-right-narrow[
<img src="data:image/png;base64,#d01.1_AI_files/figure-html/ai-1.png" width="50%" style="display: block; margin: auto;" />
]

---

## Stage 2: AI for debugging

When you're comfortable with basic R, you can use AI to:
- Identify syntax errors
- Debug error messages
- Fix common problems

--
  
.question[
  What makes a good AI debugging question?
]

--
  
Include:
- Your code
- The error message
- What you've tried
- Your expected output

---

## Stage 3: Advanced AI use

.pull-left[
In later stages, you can use AI for:
- Code generation
- Optimization suggestions
- Alternative approaches
]
--
.pull-right[
But remember to:
- Verify the output
- Understand the code
- Test the solution
- Consider alternatives
]

---

## Signs you're ready for AI tools

.pull-left[
  You can:
  - Write basic R code
  - Spot obvious errors
  - Debug simple problems
  - Read documentation
]
--
.pull-right[
  You understand:
  - Data structures
  - Function basics
  - Pipe operations
  - ggplot2 grammar
]

---

# Practical Guidance on Using AI

---
  
## Example: Appropriate AI use
  
+ Good AI question:
```r
# "Can you explain why this gives an error?
ggplot(diamonds, aes(x = carat) +
  geom_histogram()"
```

+ Less-Good AI question:

```r
# "This doesn't work"
ggplot(diamonds, aes(x = carat) +
  geom_histogram()
```

+ Even Less-Good AI question:

```r
# "histogram the diamonds data in ggplot"
```
---

# Guidelines for AI use

---
  
## When to use AI
  
.pull-left[
✅ Good times:
- Understanding complex code
- Debugging specific errors
- Learning alternative approaches
- Checking your work
- For speeding up specific tasks you can already do
]

--
.pull-right[
❌ Bad times:
- Before trying yourself
- For basic syntax
- Without understanding the output
- As a substitute for learning
- For entire assignments
]

---
  
## How to evaluate AI suggestions
  
Before using AI code:
  
- Can you explain what each line does?
- Does it follow your expectations?
- Would you be able to modify it?
- Can you spot any potential problems?
  
---
  
## Example: Evaluating AI code
  
.pull-left[
AI suggests:
```r
    diamonds_summary <- diamonds %>%
      group_by(cut) %>%
      summarize(
        avg_price = mean(price),
        count = n()
      ) %>%
      arrange(desc(avg_price))
```
]

.pull-right[
Check:
- Do you understand each function?
- Does the output match your expectations?
- Is this the best way to solve the problem?
- What assumptions does this code make?
- How would you modify it?
]

---

## Key points

- Build R foundations first
- Use AI tools gradually
- Always verify AI suggestions
- Understand before using
- Practice critical evaluation

---

# Wrapping Up...