Drexel UniversityDornsife School of Public Health

Department of Environmental
& Occupational Health

AI WORKSHOPS FOR DSPH

Workflow Enhancement

Use ChatGPT Edu, Copilot Chat, and Microsoft 365 Copilot to speed up the work you already do

Alex Quistberg, PhD, MPHAssociate Research Professor · Department of Environmental & Occupational Health

Workshop 2 · July 7, 2026

WHAT YOU'LL LEAVE WITH

Agenda & Objectives

1Spot which recurring tasks are good candidates for AI.
2Use Microsoft 365 Copilot in Word, Outlook, and Excel — with three worked examples.
3Use ChatGPT Edu and Copilot Chat to work faster.
4Build a small prompt recipe you can reuse.
5Verify, disclose, and avoid over-reliance on AI.

SECTION 01

Where AI Fits

The biggest time savings come from the repetitive middle of a task.

THE SHAPE OF A TASK

AI Takes the Middle — You Keep Both Ends

WHICH TOOL, WHEN

Three Tools, Three Jobs

Microsoft 365 Copilot

Best when the work is already in a Microsoft file — a Word doc to rewrite, an Outlook thread to summarize, an Excel sheet to analyze. Sees the file you're in; respects Microsoft 365 permissions.

Copilot Chat (EDP)

Protected web-based drafting, summarizing, and brainstorming with Enterprise Data Protection — when you don't need a reusable GPT.

ChatGPT Edu

Open-ended drafting, summarizing pasted text, file analysis, and reusable assistants. Use GPT-5.5 when it appears in your model picker; otherwise the current default.

A BORING, PRACTICAL CHOICE

Pick the Tool by Where the Work Lives

FIND YOUR CANDIDATES

What Makes a Good AI Task

CONCRETE CANDIDATES

A Week of DSPH Work, Mapped to a Tool

Role	Task you already do	Where AI helps
Research	Turn a specific aims paragraph into a lay summary for a community board; summarize reviewer comments into a response checklist	Word Copilot · ChatGPT Edu
Research	First-pass summary of 3–5 abstracts into a comparison table for a team meeting	ChatGPT Edu (paste text)
Teaching	Draft a course announcement from bullet points; turn lecture notes into a study guide; tighten rubric language	Word Copilot · Copilot Chat
Admin	Summarize a 12-message committee thread into decisions and action items; draft the reply	Outlook Copilot
Admin	Sanity-check a small spreadsheet: totals, flags, a quick trend chart	Excel Copilot

All five show up as worked examples in today's hands-on segments.

SECTION 02

Ethics & Governance

Verification, disclosure, and over-reliance — what accountability means in daily work.

WORKFLOW ETHICS

Three Habits That Keep You Safe

Verify Before It Counts

AI can fabricate facts, stats, and citations that read as authoritative. Check anything going into a real document or decision against the source — especially numbers and references.

Disclose Appropriately

For official work products, external communications, and anything research-related, note that AI assisted — per Drexel's AI policy and your unit's expectations.

Avoid Over-Reliance

Speed is the selling point and the trap. If AI does all the drafting and thinking, your skill and your voice both erode. Keep your judgment in the loop.

A CHECK YOU CAN RUN IN 60 SECONDS

Before You Hit Send

“

If you can’t open it, you can’t cite it.

A rule worth keepingTreat every AI-supplied figure as unverified until you’ve confirmed it

SECTION 03

Hands-On

Three worked examples in Office, then one real task in ChatGPT Edu. Demo files are on the workshop page.

ACTIVITY 1 · ~19 MIN

Microsoft 365 Copilot in Office

Pick the track that matches your week. Each has a demo file if you didn't bring your own — the next three slides walk through one worked example per track.

Track A — Word

Research: turn a study description into a lay summary for a community advisory board, then draft a next-steps section.

Track B — Outlook

Teaching/committee: pull the decisions and action items out of a 6-message curriculum-committee thread, then draft the reply.

Track C — Excel

Admin: summarize a seminar-attendance sheet, flag high cost-per-attendee rows with a formula, chart the trend.

WORKED EXAMPLE · TRACK A · WORD

Research: From Study Aims to Lay Summary

Demo file · Workshop2_Demo_StudyDescription.docx

Pedestrian injuries remain a leading cause of traffic death in U.S. cities, and risk is concentrated at a small share of intersections. This study will link citywide crash records with street-level imagery to identify built-environment features — crossing distance, vehicle speeds, lighting, signal timing — associated with severe pedestrian injury at signalized intersections. Results will produce a ranked list of modifiable intersection features for city transportation partners…

Step 1 · SummarizeSummarize this document in 5 bullet points for a project status update.

Step 2 · Rewrite for an audienceRewrite the first paragraph for a community advisory board — no jargon, 8th-grade reading level, keep it to 4 sentences.

Step 3 · Draft new contentDraft a short "Next Steps" section: one paragraph, three concrete steps, active voice.

Watch for: does the lay version keep the numbers accurate? That's your verify step.

WORKED EXAMPLE · TRACK B · OUTLOOK

Committee Thread: Find the Decisions, Draft the Reply

Demo file · Workshop2_Demo_EmailThread.docx (6 messages, excerpt)

M. Rivera: …proposing we move EPI 570 to fall and add the data-visualization module. Thoughts before Friday?

J. Chen: Fall works if the lab sections don't collide with BST 555. Also — who owns the syllabus update?

M. Rivera: Agreed on fall, pending the registrar check. Syllabus update: can you take it, with a July 20 draft?

+ 3 more messages…

Step 1 · SummarizeSummarize this thread. List every decision made and every open question.

Step 2 · Structure itList the action items as a table: owner, task, due date.

Step 3 · Draft the replyDraft a reply confirming I'll have the syllabus draft ready by July 20 — concise and friendly, two sentences.

Watch for: did it invent a decision that was never actually made? Threads are where AI over-summarizes.

WORKED EXAMPLE · TRACK C · EXCEL

Admin: Seminar Attendance & Spend

Demo file · Workshop2_Demo_SeminarAttendance.xlsx (30 rows, excerpt)

Month	Department	Attendees	Cost ($)
Sep 2025	EOH	34	310
Sep 2025	Epi & Biostat	51	445
Oct 2025	EOH	29	402
Oct 2025	HMP	22	295
… Sep–Jun, three departments

Step 1 · SummarizeSummarize this data. What stands out?

Step 2 · Formula in plain languageAdd a column that flags rows where cost per attendee is over $12.

Step 3 · Chart itSuggest a chart showing the attendance trend by department across the year.

Watch for: check the flag formula against one row by hand before you trust the column.

ACTIVITY 2 · ~13 MIN

ChatGPT Edu: Anatomy of a Working Prompt

Run one task you actually need this week. Structure the prompt with PTCF — here it is filled in for the lay-summary task:

Persona

You are a research coordinator at a school of public health.

Task

Write a 150-word lay summary of the study description below.

Context

Audience: a community advisory board. No jargon; they care about what changes in their neighborhood.

Format

One paragraph, 8th-grade reading level, end with a one-sentence takeaway.

WORKED EXAMPLE · CHATGPT EDU

Summarize-Then-Act: Abstracts to Team Email

Demo file · Workshop2_Demo_Abstracts.docx (3 short abstracts, excerpt)

Abstract 1. A citywide study of nighttime pedestrian crashes and street lighting upgrades…

Abstract 2. Speed-limit reductions on urban arterials and severe-injury rates…

Abstract 3. Computer-vision audit of crosswalk quality across 2,000 intersections…

One prompt, two productsSummarize each abstract as one row of a table: population, method, key finding. Then draft a 3-sentence email to my research team on what these mean for our streetlight and injury analysis.

The patternSummarize [material], then [the thing you'd do next with it].

One step takes you from raw material to a product you can nearly send. The "make it a table" move works on anything with structure.

DON'T SETTLE FOR DRAFT ONE

Iterate: Three Rounds Beats One Perfect Prompt

Round 1

[the PTCF prompt from the last slide]

Output is usable but generic — reads like anyone could have written it.

Round 2

Cut it to half the length. Make the findings column quantitative — effect sizes, not adjectives.

Tighter, more specific. Numbers now need your verification.

Round 3

Add a column rating each study's relevance to our project, 1–3, with a one-phrase justification.

Now it's doing analysis for you to react to — you accept, reject, or correct.

Then do the final edit yourself. What you changed is the part that needed your expertise.

SAVE WHAT WORKED

Build a Reusable Recipe

Once a prompt works, strip the specifics and leave placeholders. Keep these in a running note — next week you fill in blanks instead of starting over.

The templateYou are a [role]. [Task]. Context: [audience, purpose, constraints]. Format: [structure, length, tone].

Saved recipe · lay summaryYou are a research coordinator. Write a [length] lay summary of the text below. Context: audience is [group]; no jargon. Format: one paragraph, 8th-grade level, end with a one-sentence takeaway.

Saved recipe · thread triageSummarize this thread. Decisions made, open questions, and action items as a table (owner, task, due date).

WRAP-UP

What to Take Away

Microsoft 365 Copilot for work inside Office files; Copilot Chat for protected web chat; ChatGPT Edu for open drafting and reusable assistants.

The win is the first draft and the summary — you keep the judgment and both ends of the task.
Run the 60-second check before anything counts: verified, disclosed, yours.
Save the prompts that work as reusable recipes.
Before Workshop 3: save two prompt recipes you’d actually reuse. We’ll turn shared recipes into team tools.

Drexel UniversityDornsife School of Public Health

THANK YOU

Questions & Discussion

Next: Workshop 3 — Collaboration on Projects. Shared Projects, custom GPTs, and a team prompt library.

Alex Quistberg, PhD, MPHdaq26@drexel.edu

Department of Environmental & Occupational Health