Michelle Blomberg

Current research

Simulation as Assessment

A working thesis about where online assessment is going: the exam gets replaced by a task. Instead of answering questions about the work, the student does the work inside an authentic, browser-based simulation, personalized to them, and the proof of competence is the performance itself.

Call for collaborators

I am looking for subject-matter experts in engineering, physics, chemistry, and data or earth and energy science to help develop and test these simulations. If you teach or work in one of these fields and want to help shape an authentic, real-world assessment in it, I would love to talk.

Ongoing research and prototyping. This page is my working landscape of the field and my own design direction, and it changes as the work develops.

I am researching and prototyping what a summative assessment becomes once two things are true at the same time: it is customized to the individual learner, and it takes the form of doing the job rather than taking a test. The traditional final exam assumes a room, a clock, and a single correct answer. Fully online programs have none of the first two, and generative AI has quietly dissolved the third. So the question I am working on is not how to defend the old exam, it is what replaces it.

This is a continuation of my own scholarship, not a new interest. My master’s research was on personal learning environments (PLEs), the idea that learning is strongest when it is organized around the individual learner and their own network of tools, people, and goals rather than a one-size-fits-all container. Personalized, learner-customized assessment is that same principle carried into how we measure competence: if the learning is built around the person, the proof of it should be too. This work grows directly out of that research.

Goal, audience, and what I am researching

Goal

Replace the AI-cheatable final exam with a summative, capstone-style assessment a student cannot fake: an authentic professional problem, personalized to the learner, where the grade rests on the process and the reasoning rather than a final answer a chatbot could hand back.

Audience

Fully online professional and technical programs whose students never come to campus, never meet in real time, and never set foot in a physical lab, and the faculty and subject matter experts who assess them. Everything has to work remotely, in a browser. The pattern is built to travel across disciplines rather than serve one course.

What I am researching

One reusable simulation-assessment pattern: personalize the scenario to the learner, have them perform a real task, and point AI at the assessment itself, as a lab partner, a simulated client, or a scenario engine, so it becomes the thing that makes authentic assessment possible instead of the thing students cheat with. I am building a small web interactive to prove it and documenting it so a subject matter expert can co-author the domain content. If the prototype proves out, the aim is to release it as an open educational resource (OER) that other instructors and programs can adopt, adapt, and reuse, so the value is a shareable open resource, not just a portfolio demo.

The thesis in one breath

Traditional summative assessments, multiple choice and written exams, exist to certify that a student can do the work. A remote student with a text box and a chatbot can now pass those without demonstrating anything, so the certification no longer certifies. The answer is not to police AI harder. It is to change what we ask students to do: put them inside an authentic problem tuned to their level and their path, make them gather what matters, make judgment calls, justify each one, and live with the consequences the scenario returns. You cannot paste your way through a well-built simulation, because the graded thing is the reasoning, not a lookup.

Two shifts sit underneath this. First, personalization: the assessment adapts to the individual learner instead of handing everyone the same fixed test. Second, performance: the student does the job rather than answering questions about it. Turned toward the assessment this way, AI stops being the cheating vector and becomes the engine, the reviewer who pushes back, the client who reacts, the scenario that adapts in real time. That is a defensible, and honestly overdue, use of AI in a technical course. The deeper working notes for this project, including the fully worked Reading the Forces example and the brainstorm brief, live alongside this page in the project folder.

A few candidate simulations

Four of the stronger candidates, tight. All are designed to run entirely in the browser as virtual simulations a fully online student completes remotely, with no physical lab, equipment, or campus visit. Each names the real task, what the student produces, and the part AI plays. Pick the one a subject matter expert can stand behind.

Additive manufacturing and materials selection

In a browser-based build-and-test simulation, a student is handed a part with a job to do and its real constraints (load case, thermal environment, cost, quantity), then chooses a material and process and defends the tradeoffs: strength against weight against cost, build orientation and anisotropy, porosity, post-processing, and why the rejected options were rejected.

AI plays a design-review engineer who interrogates the choices, and the scene shows the consequence: the part warps, delaminates, or fails at the weak axis when the reasoning is off.

Failure analysis and root cause

Working from a digital case file on screen, the student inspects a component that failed in service plus the evidence (fracture-surface images, load history, environment, a timeline) and must diagnose the root cause and recommend a fix, showing the chain from evidence to conclusion.

AI plays a skeptical senior engineer who keeps asking what rules out fatigue, or why not a material defect, forcing reasoning from the evidence rather than a guess. Hard to fake, because it is a judgment chain over specific evidence.

Data science case

Given a messy, realistic dataset and a real question, the student frames the problem, chooses a method, catches the traps (leakage, bias, a misleading metric), and defends the analysis and its limits rather than just reporting a number.

AI plays a stakeholder or product manager who keeps asking why this model, what happens if the data shifts, and what would change your mind. The transcript of that defense is the evidence.

Engineering and technology management decision simulation

The student acts as a manager inside an evolving project, budget, and risk scenario, making a sequence of interdependent decisions with real tradeoffs while the simulation reacts and changes the state.

AI plays the stakeholders in the room, an executive, a client, a team lead, each with their own pressure. The student is judged on the quality of reasoning under uncertainty, not on reaching one right ending.

How I would build it

This is scoped to what I can actually make. The prototype is a browser-based interactive in HTML, CSS, and JavaScript, built with Claude Design and AI coding help. The scenes and the visual system are drawn in Illustrator and exported as SVG, and the parts that need to react and reason, the lab partner, the simulated client, the scenario responses, run through an AI model over its API. I design and build the interface and the interaction, and AI handles the reasoning. Any motion is light and optional, and nothing here needs a game engine or 3D modeling.

A fuller immersive or XR version, walking through a scene in three dimensions, is an aspirational long-term direction, not part of my hands-on prototype. That would take a technical collaborator and a real engine, so I treat it as a roadmap item and keep the version I build firmly in the browser.

Prior art: what the field and the institutions are doing

This is my working landscape of real scholarship and real practice, grouped by the argument it supports. The through-line: as AI erodes the validity of the traditional exam, the field is moving toward authentic, personalized, performance-based assessment, and the strongest examples are already online and browser-based, which is exactly the constraint these programs live under.

1. The problem: AI and the validity of the traditional exam

The scholarship reframing academic integrity as a validity problem, not just a cheating problem.

2. Authentic and simulation-based assessment: the alternative

Peer-reviewed work establishing authentic, performance-based, and embedded assessment, including in fully digital form.

3. Personalized, adaptive, and competency-based assessment

The move from a fixed test toward assessment that adapts to the individual learner and certifies demonstrated mastery.

4. Online, at-scale, and browser-based simulation in practice

Working examples that already deliver assessment and simulated real work fully online, no physical lab required.

5. AI in teaching and assessment at named institutions and in industry

Large-scale institutional and industry investment in AI for teaching and assessment.

← Back to Work