Artificial intelligence is rapidly invading many facets of our personal and professional lives. As I’ve often said in my workshops, I’m neither an evangelist for AI nor do I think the sky is falling. Like any new technology, use of AI comes with meaningful risks. And like any new technology, it needs guardrails to be applied safely in mental health care.
Recently a couple of our exam prep competitors buried a disclosure on their web sites, deep in the small text, noting that they’re using AI to write at least some of their exam prep content. They say that this is done with human supervision, and also note the possibility of AI making errors.
We want you to know where we stand on the use of AI in exam prep.
Our content is human-written
All of our learning content – every video and every practice question – is authored by real humans with real expertise in their relevant fields.
Even our detailed test-taking skills report, which many customers understandably think is AI-powered, is actually driven by human-written algorithms and delivers human-written feedback. I know, because I wrote those things myself.
We don’t think using AI is some kind of moral failure. For gathering research or answering some questions, it can be genuinely helpful. We don’t begrudge any examinee using all of the tools available to them to prepare for a test. But we know that AI makes important mistakes, both in basic facts and in more technical areas.
Examples of AI mistakes
As an example of a basic fact that AI often gets wrong, if you ask your favorite AI platform whether California uses the National MFT Exam, it might say yes. (California is planning on switching to the National MFT Exam sometime in 2027, but until then, uses its own California MFT Clinical Exam.)
Even if it gets that right, AI might guide you incorrectly when it comes to what's actually covered on your exam. That's exactly what happened when I asked a popular AI model about the California exam today. I asked, "What are the domains of knowledge on the California MFT Exam?" It gave me back the specific domains from the national exam instead:
As an example of a more technical error, if you ask your favorite AI platform questions about child abuse reporting, quite often you’ll get responses that are inconsistent with applicable law where you are. For example, AI incorrectly states here that therapists are required to report emotional abuse in California (emotional abuse operates on a permissive reporting standard in the state):
In this instance, the AI did go on to explain that not all instances of emotional harm rise to the level of abuse. But it didn't clarify that in California, emotional abuse itself is a permissive, and not a mandated, report.
If you relied on AI in your exam prep and it confidently presented this "Yes" to you, would you ever know it was wrong? Or would you just be left wondering why you missed so many questions on your actual test?
Why AI makes mistakes
These mistakes can happen for a variety of reasons. Often they are because the large language models (LLMs) that power AI platforms are pulling from a variety of sources, and may mush together different sets of information (especially state-specific laws and profession-specific ethics codes) without recognizing where some rules or cases are being misapplied.
But the reasons aren’t especially important. The upshot is that AI can make meaningful errors in the information it presents to you, while appearing confident and authoritative in that presentation.
Of course, human beings aren’t above making mistakes, or even appearing confident while doing so. But at High Pass Education, we have a depth of experience and expertise that our competitors just don’t have.
I’ve been teaching at the graduate level for more than 20 years, and attending licensing board meetings for about as long. I’ve served on the ethics committee for my professional association. I’ve helped draft statutory language, specific ethical standards, and guidelines for the use of technology. I don’t just know what the rules are guiding our professions, I often know why those rules were put into place, and in many instances, I had a direct role in shaping them.
In its present form, we can’t endorse using AI in your exam prep.
The quality of the information and preparation you will actually get is deeply questionable, to say nothing about the significant social and environmental impacts of AI on the world around us.
I’m not ruling out the possibility that we may use AI for some purposes in the future. Again, I’m neither an evangelist nor a doomsayer.
But it’s worth asking: If AI technology is truly beneficial for exam prep, why would our competitors seem to be trying to hide the fact that they’re using it?
