Mid-Course Check-in: What You Said, What Happens Next

ACCTG 5150 — Accounting Analytics · Summer 2026 · From Sean · Preliminary — based on the first 11 survey responses

This Class is Hard By Design

This course is new, it is hard, and it is built to challenge you. You are learning SQL, Excel, data cleaning, and how to direct an AI. All in the same term, often in the same week. This change is actually happening in industry too, so you are in good company as most practicing accountants are updating their skill sets. You are not behind, as "normal" is a constantly moving target where new things become available.

I have admittedly made the class tough and am trying to overwhelm you with lots of information to make it impossible to avoid AI. With the pressure being added by this course, it is not my expectation that you should try to figure out something for 3 hours and come away feeling demoralized. If you feel stuck, and you cannot prompt your way out of it, please reach out.

The reason for the generous extension policy is to account for the wide range of AI skills that students are bringing into the class. AI lets you explore in an afternoon what used to take days or weeks to resolve. A rapid prototype that's mostly wrong still teaches you more than an evening of staring at a blank sheet. None of those moves are cheating; they are the job.

The skill being built here isn't memorizing syntax, it isn't being a software engineer, it is building confidence that your creative thinking + AI can take you quite far. As you all enter the workforce, you are unencumbered by the history of "how it was done" and uniquely positioned to be the first AI-native workforce that pushes things forward to "how it will be."

My job is to push you past what you thought you could do and to teach you how to leverage your unique superpower as accountants into creating durable solutions that balance risk and efficiency.

Question everything!

Eleven of you have filled out the mid-class survey. Thank you. Every response below shaped this page. This is what you told me, what I'm keeping, and the options I'm weighing for the back half of the course. Treat all of it as preliminary — eleven responses is a start, not a verdict, and this page gets revised as more of you weigh in. All numbers here are class-level aggregates from an anonymous survey; nothing on this page can be traced to any individual student, including by me.

First: I need more of you to weigh in

Before I make any big changes to the course, I want to hear from more than half the class. Right now the survey is at 11 of 37. The median respondent finished in about 4 minutes. It is anonymous, ungraded, and it directly changes what I do next — this page exists because eleven people spent 4 minutes. Anonymous here means Canvas strips identities before I see anything: I get answers and counts, never who.

11
goal: 19+ responses (over half the class)

This round closes Sunday, June 14 at 11:59 PM (Mountain). I'll read everything submitted by then before deciding what changes.

Take the survey (about 4 minutes)

What you said is working — expect more of the same

The clearest signals from the survey, and what I'm doing about each: more of exactly that.

Staying, and expanding

  • The step-by-step lab walkthroughs. The most-repeated positive in the survey. Every remaining lab keeps this structure.
  • The practice tools and trainers. The SQL trainer and the in-notes practice came up repeatedly. More interactive practice is coming in the remaining weeks, not less.
  • The AI-first design. Several of you said using AI as your tutor and pair-worker is genuinely teaching you. That is the whole bet of this course, and it stays. The frustrations in the next column are mine to make easier, not yours to endure alone.
  • Webinar recordings, frequent announcements, course navigation. All named, all staying.
  • No exams. Labs and homework remain the whole grade. That isn't changing.

What you said is getting in the way

  • AI frustration. Not knowing what to ask, hitting free-tier message limits mid-assignment, losing hours to errors the AI can't untangle. Worth naming: knowing what to ask is the exact skill this course trains, and the working sessions below are aimed straight at it.
  • Long instructions. Especially after a workday — several of you are feeding them to ChatGPT for a plain-English version.
  • Not knowing where an error crept in until the grade comes back.
  • Everything is new at once. SQL, Excel automation, Python, and directing an AI — often all in the same week.
  • Speed versus retention. Moving fast can feel like nothing sticks, and double-checking what the AI gives you takes time the schedule doesn't seem to grant. For the record: verifying AI output instead of trusting it is not slowness. It is the discipline this field is hiring for.

Options on the table

Here's what I'm considering in response. The survey hitting 19+ responses is what turns "considering" into "scheduling" — I want the changes to reflect the whole class, not just the eleven voices I've heard so far.

Live AI working sessions

A few live sessions where you bring any AI problem or question — from this class or not — and we work on it together, live. Stuck prompts, weird errors, "why does it keep doing this," all fair game.

Office hours, on your schedule

Message me on Canvas and we'll set up a time that works. This is available right now, not pending the survey. If you're losing hours to something, that's the signal to send the message.

Short "what this assignment wants" videos

A couple of you asked for quick videos explaining what each lab and homework is really asking for, and how to point your AI at it. Under consideration for the remaining weeks.

Worked solutions after the deadline

Seeing the full worked lab after it closes, so you can find exactly where your version went sideways. Under consideration, starting with the most recent labs.

Three things you don't have to wait for. If you are stuck, reach out via Canvas today. Several survey responses described losing hours to a single error. A Canvas message costs you two minutes and usually saves you the evening. The Ask AI button on the slides and trainers is built for exactly the "I don't know what to ask" moment — it already sees the page you're on, so you can start with nothing more than "I'm stuck." And if free-tier message caps are cutting your work sessions short, there is now a Gemini free tier running on my own API key. Message me on Canvas and I'll get you set up.

What the class looks like from my side

A few aggregate numbers, shared in the same spirit as the webinar stats in the weekly announcements — anonymized, and never factored into grades.

Everyone who responded is using AI

ChatGPT
9
Claude
5
Microsoft Copilot
3
Gemini
3
GitHub Copilot
2
Local / open-source
1
Of 11 survey respondents, every one is using at least one AI tool. If you're hitting free-tier limits, two immediate moves: switch models for the rest of the day (a cap on ChatGPT doesn't bind Claude or Gemini, and every lab works with any of them), or use the new Gemini free tier running on my own API key — message me on Canvas and I'll get you connected.

You're putting in real hours

Under 3 hrs/wk
0
3–5 hrs/wk
3
5–7 hrs/wk
4
7–10 hrs/wk
4
Over 10 hrs/wk
0
Every respondent landed between 3 and 10 hours a week, around the expected range for a 3-credit summer course. If your weeks are running well past 10, reach out — that usually means one fixable thing is eating your time.

When you message me, you get an answer

"Reach out via Canvas" is only good advice if the inbox actually answers. Here is how that has gone so far this term:

20 of 37
students have already messaged me on Canvas this term
~9 hours
median time from your message to my reply (mean is about 10)
28 of 30
message threads answered within 24 hours

Measured from the first message in each thread to my reply, across the whole term so far. More than half the class has already used this door. It is open to the rest of you too.

The one thing to do today

It takes about 4 minutes, it's anonymous and ungraded, and it decides what the back half of this course looks like. This round closes Sunday, June 14 at 11:59 PM (Mountain).

Take the mid-class survey