Before You Adopt an AI Simulation Platform

With the recent AI boom, many schools and universities are feeling pressure to stay "up to date" with the latest tech. I'm a proponent of lifelong learning and I value innovation, but there's a big gap between adopting technology because it's new and thoughtfully innovating within your program. Whether a new system becomes one more headache to learn or a real improvement to your curriculum is rarely about the software alone. It's about the plan.

It's also about the partner. AI tools change fast, and the market is in constant flux. A single model update can change how a carefully built scenario behaves overnight. A program that wants to weave simulated role-play throughout its curriculum needs a vendor it can trust to:

• share its pedagogical commitments

• be transparent about changes

• grow with the program over time

The questions below are as much about finding that shared vision as they are about checking features.

1. How, specifically, do you plan to use it?

Adopting a new platform without a plan isn't automatically a good thing, and it can quietly crowd out live practice. Asking instructors to shoehorn in even more content doesn't make a program stronger. What will this replace? How does it fit with the frameworks and approaches you already use? Before you sit through any demo, consider:

• which courses, and which stage of the program, will use it

• what it adds to live practice

• how instructors plan to integrate it into their classes

• who designs the scenarios and who reviews transcripts

• how you'll know it's working

2. Does the simulated client behave like a real client for a whole session?

Anyone who has spent more than five minutes with an LLM has been told their ideas are brilliant. That has less to do with the genius of every user than with how these systems are built: LLMs are designed to be agreeable. That may be fine for other uses, but most of my actual clients aren't so encouraging, optimistic, and eager to please (and if they are, that's a clinical matter of its own). Platforms that aren't designed by and for clinicians often miss this and its impact on training. Their clients open up and wrap up too easily. Students come away with procedural fluency and false confidence instead of clinical judgment and cognitive complexity.

As you try different platforms, look for consistency over a full session. Many AI clients drift toward resolution or lose the thread of their own story well before a clinical hour is up, let alone across multiple sessions. Stress-test for this. Run a full-length session yourself and include a middling intervention. Does the client hold onto its history? Does it make the student work for change?

Red flag: Every session ends in a tidy resolution.

3. Is it built for the ethical realities of clinical training?

For very good reasons, most off-the-shelf AI chatbots have safety guardrails built in to prevent misuse and protect users. Those guardrails help everyday users, but they can get in the way of clinical training. Counselors in training need practice assessing risk, working through ethical dilemmas, and staying present with intense material. If a simulated client breaks character to post a hotline number the moment suicide comes up, students never get to practice the assessment. Does the system you're evaluating make room for that work?

The same goes for how a platform handles your students' data. Ask about confidentiality, where data is stored, who can delete it, and whether student work will be used to train future models. How a vendor answers these questions tells you a lot about what kind of partner they'll be. Do they take your concerns as seriously as you do?

Red flag: "We're built on [brand-name AI model], so it's safe."

4. Who does the client sound like?

Look for real range across age, culture, communication style, and relational dynamics, without stereotyping. Test the platform with populations outside your students' placements, including minors if you train school counselors. AI has a track record of falling into unhelpful and even offensive stereotypes, in ways that run contrary to our ethical standards of care. Does this platform reinforce existing bias, or teach your students that every client from a given group will act a certain way? Make sure it can reflect the diversity of clinical presentations your students will actually face.

5. Does it deepen relational learning, or replace it?

This one is close to my heart as a counselor and an educator. Technology that replaces human connection risks isolating us and our students even further. Counseling is a relational profession, and training for it should be relational too. Most AI practice defaults to a solitary setup: one student, one screen, one bot. The research, and most of our own experience, suggests the real growth comes from reflection and collaboration, not from AI feedback alone. So ask whether the platform supports:

• co-therapy and peer consultation mid-session

• collaborative reflection

• transcripts students can bring to a supervision-style debrief

Or does it reinforce an individualistic, isolating experience of education and clinical work? Experience alone doesn't teach. Learning comes from reflecting on experience, and that reflection goes deeper when it's shared.

Red flag: The platform's selling point is that students "don't need a partner to practice."

6. Can faculty actually run it?

This is the most practical question, about both the platform and your context. A brand-new iPhone is fun and exciting, but it's an expensive paperweight if the person holding it isn't interested in using it or able to. If you're the one shopping for an AI platform, you're probably the tech-forward outlier in your department. Your colleagues may not share your experience or comfort with digital tools, so review your options with them in mind. Look for:

• scenario building that doesn't require coding, easy transcript review, and a good fit with your LMS

• a low barrier to entry for instructors and students, so the platform doesn't favor people already comfortable with AI

• reliable performance on ordinary devices and internet connections

• pricing that works for a full cohort, not just a pilot

A third path

Integrating AI into your curriculum can be a powerful way to help your program innovate and better prepare students for their future clients. It can also be one more drain on time, effort, and limited budgets. Neither uncritical adoption nor fear-based avoidance is the answer. I've found a third path helpful for me and my students, one where AI returns students to each other rather than pulling them further into a screen. When I build AI role-play into lessons that benefit from hands-on practice, and then have students process those experiences together, the classroom comes alive. It's also a lot more fun.