Tests Measure Recall. Simulations Measure Judgment.

A professor assigns a 20-page PDF case study. Most students don't read it. ChatGPT can summarize it in 15 seconds. The exam is multiple choice. The student passes. But can they make the right call when the CFO is in the room, the numbers are ambiguous, and the clock is running?

That gap between what we test and what actually matters is the central crisis of modern assessment, in both higher education and corporate training. Research by Burchfield and Sappington found that fewer than one in three college students complete assigned readings, and compliance rates have been declining for decades. We built a testing infrastructure on the assumption that students read. Most don't.

The gap between what we test and what matters

Multiple-choice exams and written essays have one thing in common: they reward the ability to retrieve or paraphrase information. They don't tell you whether a learner can reason through a genuine dilemma.

That distinction has always mattered. It matters even more now. Generative AI makes it trivial to produce a polished essay on any topic without reading a single source. Students who outsource their thinking to ChatGPT can still score well on recall-based assessments. The tool does the synthesis and the learner watches.

The skills that actually determine professional outcomes — decision-making under pressure, weighing incomplete information, navigating competing priorities, accepting the consequences of a wrong call — are almost never assessed. We train people in a consequence-free vacuum and then wonder why they freeze in real situations. A compliance training module with a 100% pass rate means nothing if the employee can't act when something actually goes wrong. The test was gameable. Reality isn't.

We covered why this matters for the essay format in a previous post, The Essay Trap: Why Decision-Based Assessments Beat AI Cheating, and dug into the detection arms race in The AI Cheating Problem Isn't About Detection. It's About Assessment Design. This post goes further: the fix isn't a better detector. It's a different kind of test.

Why decision-based assessment changes the equation

A decision-forcing scenario works differently. It puts the learner inside a problem before they have all the answers. Every choice has consequences that unfold in real time. There's no rubric to reverse-engineer. There's no ChatGPT-generated answer to paste in, because the simulation doesn't ask for an opinion. It demands a decision, right now, with the information on the screen.

That's not just a motivational trick. It's a fundamentally different cognitive demand. Bloom's Taxonomy places analysis, evaluation, and creation at the top of the learning hierarchy for a reason. Most assessments never get there. Decision-based simulations can't avoid them.

LiveCase is built around this principle. Learners step into branching narratives where every choice matters and no two paths are identical. Comparing notes with a classmate or asking an AI to replay the case for you doesn't help. The scenario has already moved on.

AI handles grading and scoring at scale — delivering instant, rubric-aligned feedback on qualitative decisions without adding a single hour to an instructor's workload. And performance analytics don't just record pass or fail. They map exactly where in the decision tree each learner struggled, so the next conversation can close actual gaps.

The result? LiveCase cases achieve 92% participant engagement rates, compared to the 12% completion rates typical of passive e-learning content. That's the difference between making people click through slides and making them think.

What simulation-based assessment looks like

Picture a learner who opens their assignment not to a PDF, but to a chat interface. They've been assigned the role of interim CEO. A CFO named Laura sends them a message: "We can't delay the launch again. What's your call?" A sales director contradicts her in the next message. A board member wants numbers the learner doesn't have yet. There's a timer.

That's what LiveCase delivers. Virtual characters provide partial information in a familiar chat format. Branching narratives mean 30 learners in the same cohort can compare experiences after the fact without having walked the same path. Multi-role structures let a full team simulate organizational dynamics, with different learners playing the CEO, CFO, and operations lead simultaneously.

When it's done, the host dashboard has logged every decision. The instructor doesn't walk into the debrief blind. They walk in knowing exactly who chose what, at what moment, and what consequences followed. That's a different kind of teaching conversation entirely.

Browse the LiveCase catalogue to see ready-to-run cases across leadership, crisis management, ethics, M&A strategy, and negotiation. If you're new to the format, start with What Is a Case Simulation? for a broader overview of how interactive scenarios replace static case studies.

The ROI: from completion metrics to real competence

Here's the corporate L&D version of the problem. Your compliance training shows 98% completion. Great. But completion is a proxy metric at best. It tells you the module was clicked through. It tells you nothing about whether the employee can recognize a compliance breach under time pressure or escalate appropriately in a real situation. The module was designed to be finished, not to be hard.

The higher education version is different but equally broken. An AI-written essay gets submitted. It looks fine. It demonstrates nothing about the student who submitted it. You grade the output of a language model and call it a student assessment.

LiveCase now has 19 cases published on Harvard Business Impact. Nine of them carry the bestseller label. That's nearly one in two — the highest ratio of any simulation provider on the platform. These aren't experimental projects. They're peer-tested, published, and running in classrooms and corporate programs right now.

The AI Case Authoring Studio removes the final barrier: cost and effort. Authoring is free, and you only pay when you deliver. Starting at $1.50 per participant with no subscription or platform fee, there's no reason not to start.

How to start building simulation-based assessments

There are three ways in, depending on how much time you have and how much control you want.

Pathway 1: Co-Create with AI. You describe the scenario, the role, the stakes, and the learning objective. The AI generates roughly 80% of the simulation blueprint — characters, dialogue, decision points, branching consequences. You review, adjust, and publish. Professors who have never built a simulation before have shipped their first case in an afternoon. Here's a step-by-step guide to doing it in 30 minutes.

Pathway 2: Start From Scratch. Full authoring control in the browser. Build the narrative beat by beat, set your own branching logic, write every character line yourself. No code required.

Pathway 3: Studio Services. LiveCase's team builds the simulation for you. White-glove production, expert instructional design, delivery-ready output. This is how the INSEAD and Harvard Business Impact cases were created — with professors shaping the content and the Studio team handling execution. Visit Studio Services for details.

Every case you build is your intellectual property. Keep it private, share it with your institution, or publish it through LiveCase's distribution partners — including Harvard Business Impact, Ivey Publishing, and The Case Centre — and earn royalties when others run it. We covered the publishing success in more depth in 9 LiveCase Bestsellers on Harvard Business Impact.

Build your first simulation today

The AI Case Authoring Studio is free to access at livecase.com. No credit card required. You can have a working simulation ready for your next session in under an hour.

If you would rather start with something already built and peer-reviewed, the catalogue has ready-to-run cases — including those published on Harvard Business Impact that you can deploy immediately.

If your organisation needs something fully custom, Studio Services is the path. Talk to the team, describe your learning objective, and they will handle the rest.

Tests measure recall. Simulations measure judgment. The tools to close that gap exist right now — and the first one costs nothing to build.

FAQ

What is decision-based assessment?

Decision-based assessment evaluates learners by placing them in realistic scenarios where they must make choices with consequences, rather than selecting answers from a list or summarizing information. It tests judgment, prioritisation, and reasoning under pressure — the skills that traditional exams typically miss.

How do AI simulations prevent cheating?

AI simulations are resistant to outsourcing because they require real-time decisions that branch uniquely based on each choice made. A learner cannot ask ChatGPT to complete a simulation the way they might draft an essay, because the narrative state changes with every response and no two paths through the simulation are identical.

How much does LiveCase cost?

Authoring and previewing on LiveCase is free. Delivery is priced per participant, starting at $1.50 per learner per run, with no subscription or platform fee. You can build and refine your simulation at no cost and only pay when you run it with a live cohort.

Do I need coding skills to build a simulation?

No coding is required at any stage. You can co-create with AI in the browser, author from scratch using the visual editor, or work with LiveCase Studio Services to have the team build the simulation for you. Learners join via a link, QR code, or LMS integration.

Can I monetize the cases I create?

Yes. Every case you build on LiveCase is your intellectual property. You can publish it through LiveCase's distribution partners including Harvard Business Impact, Ivey Publishing, and The Case Centre — and earn royalties each time another educator or organisation runs it.