A University of Arizona · BIO5 Institute
PH&AI Summer School · 2nd Edition
AI Fluency Track · Deep Dive

Prompt
Engineering,
on purpose.

Tyson Swetnam, PhD
Associate Professor · BIO5 Institute
Grand Challenges Research Building
Tucson, AZ · June 2026
APH&AI · AI Fluency
02 / Outcome
What You'll Leave With

Four things, transferable across every chat window you open.

This deck mirrors intro-gpt/prompts/. Pull it up alongside — every section in the slides has a deeper write-up there.

  • 01
    Fundamentals. How AI models actually process and respond to prompts.
  • 02
    Modern features. Document upload, web search, multi-modal inputs.
  • 03
    Best practices. Structured approaches to writing effective prompts.
  • 04
    Advanced techniques. Context management, chaining, custom instructions.
APH&AI · AI Fluency
Part 01 · Mental Model
Part 01

What the model
actually sees.

Five stages from your keystrokes to its first token. Skip this and you'll spend the rest of your career arguing with a black box.

APH&AI · AI Fluency
04 / The Pipeline
How AI Models Process Your Input

Five stages between your keystrokes and the first generated token.

01
Tokenization.

Your prompt is broken into smaller units — tokens. Whitespace, punctuation, sub-words.

02
Context assembly.

Uploaded documents and prior conversation are stitched in alongside your message.

03
Attention.

The model weighs which parts of that context matter for the next token. This is where prompt phrasing pays off.

04
Generation.

Response is produced token by token, conditioned on everything seen so far.

05
Formatting.

Output is structured according to your specifications — markdown, JSON, code blocks.

Most "the AI didn't listen" complaints are actually stage 03 problems: the relevant instruction was buried in a wall of context, and the model attended to something else.

APH&AI · AI Fluency
05 / Tooling 2026
Core Features of Today's AI Tools

The picture in mid-2026 — pick the surface that matches the job.

Feature Claude ChatGPT Gemini NotebookLM Copilot
Document upload PDFs, text, code PDFs, images, data PDFs, images, GDrive PDFs, Google Docs PDFs, OneDrive
Web search via MCP Yes Yes Yes Yes
Context window 200K 128K 2M Document-based 128K
File analysis Yes Yes Yes Deep analysis Yes
Code execution Yes (MCP) Yes Yes No Yes

Context-window size is rarely the bottleneck — placement and phrasing inside it usually are. A 2M-token window does not exempt you from CRAFT.

APH&AI · AI Fluency
06 / Anatomy
The Foundation · Clear Instructions

Same task, three drafts. Watch the prompt grow a spine.

Basic
# Basic
"Summarize this
research paper
in 3 bullet points"

A request. No role. No structure. You can't predict what you'll get.

Better
# Better
"As a research
scientist, summarize
the key findings
from this paper in
3 bullets, focusing
on methodology and
results"

Adds a role and a focus.

Best
# Best
"You are a research
scientist reviewing
for a journal.
Summarize the PDF
in 3 bullets:
1. Q & hypothesis
2. Methods & n
3. Findings & limits
Format as a list."

Role · audience · scope · format. Reproducible.

APH&AI · AI Fluency
Part 02 · The CRAFT Framework
Part 02

C·R·A·F·T

Context · Role · Action · Format · Tone. Five fields. One mnemonic. The fastest way to make a prompt reproducible.

APH&AI · AI Fluency
08 / CRAFT · 01
C.
01 · Context

Set the scene.

Provide background information. What project is this? Who is the audience? What's the upstream goal that this prompt serves?

In practice

"I'm preparing a grant proposal for NSF funding on AI in education…"

APH&AI · AI Fluency
09 / CRAFT · 02
R.
02 · Role

Define who the AI is.

A role anchors the model to a register, a vocabulary, and a set of priors. "Grant writer" produces different prose than "tweet writer", even with the same task.

In practice

"Act as an experienced grant writer and education researcher…"

APH&AI · AI Fluency
10 / CRAFT · 03
A.
03 · Action

Specify the verb.

One unambiguous task. "Review" is not "rewrite". "Summarize" is not "expand". If you need three actions, number them.

In practice

"Review my draft introduction and suggest improvements…"

APH&AI · AI Fluency
11 / CRAFT · 04
F.
04 · Format

Describe the shape.

Markdown headers, bulleted lists, a comparison table, a JSON object, tracked-changes. The model will pick a default if you don't — and you usually won't like it.

In practice

"Provide feedback as tracked changes with explanations…"

APH&AI · AI Fluency
12 / CRAFT · 05
T.
05 · Tone

Indicate style and voice.

Professional · constructive · skeptical · plain-English-for-a-policymaker. Tone is the one CRAFT field most people skip — and it's the difference between usable and rewriting from scratch.

In practice

"Professional, constructive, and encouraging."

APH&AI · AI Fluency
13 / CRAFT in Full
CRAFT · Worked Example

All five fields, one prompt, copy-able.

Context: I'm preparing a grant proposal
   for NSF funding on AI in education.

Role:    Act as an experienced grant
   writer and education researcher.

Action:  Review my draft introduction
   and suggest improvements.

Format:  Provide feedback as tracked
   changes with explanations.

Tone:    Professional, constructive,
   and encouraging.
Why it works

Every field constrains a different degree of freedom. The model has nowhere to drift.

Save it

A reusable CRAFT block becomes a custom instruction, a Project, or a prompt template.

APH&AI · AI Fluency
Part 03 · Advanced Techniques
Part 03

Past the
single prompt.

System instructions · few-shot · chaining · multi-modal · web search. The vocabulary you need to recognize, even if you only use three of them this month.

APH&AI · AI Fluency
15 / Technique 01
Technique 01

Custom instructions: CRAFT, but persistent.

# Project Context
I'm a data scientist working on
machine learning projects.
Always provide Python code examples
using scikit-learn and pandas.
Include docstrings and type hints
in all code.

# Response Preferences
- Be concise but thorough
- Explain complex concepts with analogies
- Always cite sources when making
  factual claims
Where it lives
Claude
System Instructions
Gemini
Custom Instructions
ChatGPT
Custom GPTs or Projects
Why bother

It becomes a global rule applied to every prompt that follows — your defaults move up to the platform layer, not the chat.

APH&AI · AI Fluency
16 / Technique 02
Technique 02 — Few-shot learning

Show, don't tell. Three labeled examples beat a paragraph of instructions.

I need to classify customer feedback.
Here are examples:

"The product arrived damaged"
   → Category: Shipping Issue
"Can't log into my account"
   → Category: Technical Support
"Love the new features!"
   → Category: Positive Feedback

Now classify these:
1. "The app keeps crashing on startup"
2. "Best purchase I've made this year"
3. "Package was left in the rain"
Rule of thumb

3–5 examples is the sweet spot. Cover the corners of the label space, not just the easy cases.

Public-health twist

Few-shot is how you teach a model your local ICD coding shorthand, your survey response taxonomy, or your team's review rubric — without fine-tuning.

APH&AI · AI Fluency
17 / Technique 03
Technique 03 — Prompt chaining

Build complex outputs through a sequence — not one mega-prompt.

Start broad
prompt 01
"Outline a research paper on sustainable AI."
Zoom in
prompt 02
"Expand section 3 on energy-efficient training methods."
Refine
prompt 03
"Add citations and make the tone more academic."
Polish
prompt 04
"Format according to IEEE standards."

Each turn corrects course. You're trading one undirected lottery ticket for four checkpoints where you can intervene.

APH&AI · AI Fluency
18 / Techniques 04 + 05
Combining inputs & live retrieval

Multi-modal prompting and web search — the two cheat codes most people leave on the table.

Technique 04 — Multi-modal
I've uploaded:
1. A screenshot of my dashboard
2. The underlying data in CSV
3. Our brand guidelines PDF

Create a redesigned dashboard:
- improve data viz
- adhere to our brand colors
- highlight the data-dict KPIs

A single prompt that sees pixels, rows, and rules at once.

Technique 05 — Web search
Search for the latest research on
the public health benefits of
vaccination published in 2024.

Focus on:
- Top conferences (AHA, ASPPH)
- mRNA, Bird Flu, COVID

Summarize the top 5 with links.

Pair retrieval with a citation requirement — or you get confident summaries with no source trail.

APH&AI · AI Fluency
Part 04 · Common Pitfalls
Part 04

Four ways
to fail.

The mistakes that show up in every workshop, every cohort, every project — and the small rewrites that fix them.

APH&AI · AI Fluency
20 / Pitfalls 01–02
The most common rewrites

Vague instructions & information overload.

Pitfall 01 — Vague instructions
Poor

"Make this better."

Better

"Improve this abstract by making it more concise (under 250 words), adding keywords, and ensuring it follows the journal's structure: background, methods, results, conclusions."

Pitfall 02 — Information overload
Poor

Uploading 50 documents without guidance.

Better

"Focus on documents 1–3 which contain the methodology. Ignore the appendices."

APH&AI · AI Fluency
21 / Pitfalls 03–04
The other two everyone makes

Assuming knowledge & leaving the format up to the model.

Pitfall 03 — Assuming knowledge
Poor

"Fix the usual issues."

Better

"Check for: passive voice, sentences over 25 words, undefined acronyms, and missing Oxford commas."

Pitfall 04 — No output format
Poor

"Summarize this."

Better

"Create an executive summary with: a 3-sentence overview, 5 key points as bullets, 1 paragraph on implications, formatted with markdown headers."

APH&AI · AI Fluency
22 / Quick Reference
Tape This Above Your Monitor

Nine boxes. Run them before you hit return.

[ ]
Clear objective.
What do you want to achieve?
[ ]
Context provided.
Background included?
[ ]
Role defined.
Who should the AI act as?
[ ]
Specific action.
Exact task described?
[ ]
Output format.
Structure specified?
[ ]
Examples given.
For complex tasks?
[ ]
Constraints noted.
Length, style, content limits?
[ ]
Documents referenced.
If using uploads?
[ ]
Follow-up planned.
For iterative improvement?
APH&AI · AI Fluency
23 / Resources
Bookmarks

Everything from today, plus the deeper write-ups.

This Page
tyson-swetnam.github.io/intro-gpt/prompts/
Full Workshop
tyson-swetnam.github.io/intro-gpt/
Anthropic
docs.anthropic.com/claude/docs/prompt-engineering
OpenAI
platform.openai.com/docs/guides/prompt-engineering
Google
ai.google.dev/gemini-api/docs/prompting-strategies
Learn Prompting
learnprompting.org
APH&AI · AI Fluency
End · Q&A
Open the lab notebook

Now you
try one.

Pull up a chat window. Pick a task from your week. Write it CRAFT-style. Compare it to the version you'd have typed an hour ago.

Tyson Swetnam, PhD
tswetnam@arizona.edu · @tswetnam
BIO5 Institute
University of Arizona