The core pain in intro accounting hasn't changed. We want students reasoning like accountants by the end of the semester, not just executing procedures. But the reasoning depends on vocabulary they don't have on day one. So the usual sequence is mechanics first, judgment later.

We've been running the sequence backwards in our own classrooms and want to share what we have learned.

Problem first, terminology second

The class opens with a situation students already know how to reason about, no accounting required. For example, a friend wants to borrow money to start a business, or a group of friends takes a trip and must split the costs. No accounting terms have entered the conversation yet.

For fifteen minutes, students reason through the situation using the intuition they already brought to class. What's the risk? Who benefits? What information would you want before deciding? Then we introduce the accounting terminology for what they just did. Cost allocation has a formal definition, but it also has a shape they now recognize because they did it. The same holds for accruals, materiality, or whatever concept the chapter covers.

Students spend the opening of class thinking, not decoding. The terminology sticks because it names something they have already reasoned out for themselves.

Why now

Problem-based learning has been an effective teaching strategy for decades. The pattern is durable: students who wrestle with an authentic problem before instruction retain more, transfer better to new situations, and stay more engaged. What's new is that AI has sharpened the argument.

Anthropic released a 2025 study of about 574,000 real student conversations with its AI. Roughly half of the time, students were asking it to produce the answer rather than help them get to one. An MIT Media Lab study the same year (Kosmyna et al., 2025) put EEG caps on students writing essays with an LLM, with search, or with no help. The more assistance students had, the less their brains lit up.

This matches what shows up in professional AI work. AI is a magnifier of the person using it, not a substitute for them. In the hands of someone who already understands the problem, it accelerates good work. Without that grounding, it produces convincing slop. If we teach students to reach for AI before they can think through a problem themselves, we're just training them to produce slop faster. We want to teach them to work the problem first, so AI becomes a magnifier of what they already understand.

Meanwhile, AI adoption in accounting firms roughly doubled in one year. The tasks most exposed to automation are the ones we tend to drill in the intro course. If we spend a semester training students to do what AI already does, we're preparing them for a version of the job that continues to shrink.

Biases and AI in the room

Two other pieces round this out. The first is cognitive bias: anchoring, confirmation bias, availability, and sunk cost. A nineteen-year-old can grasp these in a class period, and once they can name them, they can spot them in a case, news story, or even their own reasoning. Several of our cases lean deliberately into a specific bias and ask students to identify the bias after they've already fallen for it.

The second is AI in the classroom. Some of our cases have students work the problem themselves, then explicitly compare their reasoning to what AI produces. AI makes plausible-sounding mistakes on accounting problems more often than students expect. Watching those mistakes happen in class has been some of the best teaching we've done in a long time. It builds AI literacy and accounting judgment simultaneously.

Where this lives

You can find these cases and more on edmondshub.com. The cases are the heart of what we teach. Additionally, the site includes a feed of current articles in the press tagged to specific chapters and learning objectives, with teaching descriptions and discussion prompts ready to go. Whatever you are teaching this week, chances are there is a story on the feed that ties to it.

The site is free, open to anyone, no sign-in required to browse. Check it out, and if you have any thoughts or ideas, we would love to hear your feedback.