Carrie Champ Morera

I am a director of product for a psychological assessment publishing company.

I did. I moved to Austin about five months ago from Florida.

I am loving it. I love the weather, love the people here. It's very chill, very relaxed, more so compared to Florida. I enjoyed Florida too. I'm happy that I'm moving from a climate that's very similar to where I was before.

No, I haven't, but Florida summers are very, very hot and humid and I understand maybe it's a little bit less humid here.

As the director of product, I oversee our product development of psychological assessment products. We have a whole array of products from clinical, neuropsychological assessments. We have educational assessments and organizational assessments, career assessments, all different types of assessments that really help people in a wide variety of situations and occupations that they're in. I oversee the development of the process from the beginning when we acquire the products all the way through the end when we release them and also manage them as well.

I manage more of the content side and I do have my hands in the tech side as well. The team of psychologists that I have, they all have a different type of expertise, and so they are our subject matter experts. We utilize them along with some of the authors of the products that we develop to really build up the content in that area of specificity. So we do the item development. We do the content development. It's really the meat of the product and coming up with the idea. I work with those teams and I closely supervise them and we collaborate.

Then once we have our idea developed, we work really close in conjunction with our technology team to really bring that product to life. We've really moved over the last few years from more of a print type of a publishing company developing psychological assessments to digital. We're doing more and more digital, and we have worked more and more closely every day with tech. So those lines are really blending.

The products are for a wide variety of psychologists and clinicians. They could be psychologists who work in schools, psychologists who work in hospitals, agencies. We also have those that work in hospital settings, private settings, doing private evaluations. It's really for those trained professionals that are doing evaluations on either children, adolescents, adults.

Sure. They're all standardized assessment products, and they can take up to several years to develop. Anywhere from six months, I think, was the quickest we got out to the gate, up to several years. So they do standardization, lots of data collection. They're norm-referenced. They are census-matched. There is a lot of development that goes into the products so that professionals are using these. They're not just quick assessment tools that you would just go out there and find free for the internet. They're very reliable. They're very valid. They're very scientific-based, and we really rely on that. Our customers rely on that and trust us to develop those trustworthy products.

Yes, psychoanalytical tools, psychological tools. For example, we have one called the RIAS, Reynolds Intellectual Assessment Scales, that assesses cognitive ability in individuals from children all the way up through adults. We also have personality assessments — for example, the Personality Assessment Inventory, which we have for adults ages 18 and up. We also have an adolescent version of that one as well. And then on the educational side, we also have standardized assessments that look at academics, including mathematics, writing, reading skills. Those are all standardized and used in the school settings, a lot so by school psychologists. They're used to diagnose, for example, learning disabilities, so that way they can determine if children qualify to receive educational services, such as individual education plans.

Yes, we are, and we're really excited about doing it. Also a little nervous at the same time in terms of: are people going to use this? Are people going to adopt this? One of the challenges that the psychologists out there in the field face is getting through the assessments in terms of report writing — putting all of that data together, putting together comprehensive reports that can be used. That can take hours and hours, right? Maybe up to 20 or so hours. What we wanted to do is develop a product that would help cut that time drastically, maybe cut it in half. So we've developed a report writer, which incorporates AI into it to help speed up that process.

What the psychologists then will do with this new product is they will upload into the system all of the test scores and the background information, the data that they have — they'll have bits and pieces. Maybe they will take part of an assessment protocol and upload it. Maybe they'll take their notes from a Word document or even their handwritten notes, and they will get that into our system. What our system will do is it will interpret that information and put together a report. It will come out very nicely in sections. The psychologists, when they go into the system, can choose the areas that they want the report to provide for them. If they want a referral section, a summary section, a cognitive assessment section, an academic section, they can choose what they want and then it will generate a report.

We have AI integrated into it, and we have some guardrails in place. Part of what we've built and what we've really structured is the actual interpretation piece. When the psychologists are giving the scores, we have certain tests that it is compatible with, so that when the report is generated, it will always say that this score is in this range and means this thing. We have those particular guardrails. And then where the more generative AI piece comes into it is more of the summary and the background information and the flow of it — to kind of tie it all together and take that data that already has been standardized and put it together in a piece that is really comprehensive and summarizes for the clinician the results.

The other exciting thing that AI has really been able to help with is in terms of generating interventions and recommendations. That is an area where psychologists may spend lots of time going through lots of different resources, or checking with their colleagues, doing lots of independent research to figure out, "Oh, what intervention works well for someone who is struggling with basic reading skills?" or "What intervention works well for a child who is really, really depressed and maybe not wanting to come to school?" What the AI has done in our report writer is take a lot of recommendations and put them into the report, saving the psychologist lots of time.

I can't really disclose that — that's proprietary information in terms of the specific models that we are using. But what I can share is that we have a closed model that we've developed internally in-house that we are using, and then in combination with a large language generative AI model.

Yes, that is correct.

The data is all randomized, it's secure, we don't save it — everything is in the customer's control. So once they run that report, we don't keep that data on our end. Once it's done, it's done.

That is what we recommend. We even have on the reports that are generated some clauses about making sure that they read through the reports, because it is always the psychologist who is putting their name on it. They are the ones who are responsible for the content in that report. So we want them to go through and reread everything that is in the report and put their final touches on it, tweak things a little bit here and there.

I will say that the pieces of the report that we are really, really confident about are the actual scores and the interpretations, because we have a lot of guardrails in place. We're feeling that, okay, that part is not going to provide hallucinations, if you will. It's more of that generative piece that comes out in the summary, the conclusions, maybe even some of the recommendations, where it might not interpret things the exact way the psychologist would actually interpret it, knowing that full client and everything about them. So it would be their responsibility then to take that information, read through it, and put it into the context that makes sense with all of the data that they have.

Yes. It saves a lot of time, and part of it too is, you are going to essentially get kind of what you put into it. So you want to provide rich information, enough information, in terms of the prompts or how you're setting it up for that to be generated. Otherwise the system is going to guess and just put words and phrases together. But if they have more detailed information about the client to begin with, then it will have more information to generate and pull from.

I think it is a really challenging landscape out there. On the one hand, there's a real need for people to receive psychological services and they aren't getting the services that they need. Wait lists are really long. It can be expensive. There is a shortage nationally of therapists that are available. And people are needing help at the moment, so they're turning to these resources and using AI to help them. On the one hand, it can be a quick, "easy" solution to help and maybe just to get through a little problem.

On the other hand, it can be challenging, because individuals then are relying on these sources to really get them through the day. They're not always giving them the right information. And they're also maybe agreeing with them and affirming them — they're not challenging them. The role of a therapist is really to challenge the client in terms of their thinking and to help them grow and develop and heal. That is not something that an AI model is capable of doing. It is going to feed that person's ego, tell them what they want to hear, agree with them, set them down a certain path. So it can be dangerous. I think there certainly needs to be some more guardrails in place.

I would also say that if there are people out there using it and then they do go to therapy and they're working with their therapist, I would encourage them to have open conversations with their therapists about it. I'm sure there are plenty of clients out there that are going to therapy but then also using AI in the interim. If they are doing that, I would encourage them to have that conversation and put some guardrails in there about how it can be helpful and how it may not, and what things you should work with your therapist directly on.

Yeah. I use ChatGPT, I use Claude. I've just been dabbling in them. For work, I've been using them — lots of people have been using this for emails now, just as a sounding board to help with that. I've been using it to help with documentation. We document all of our processes, and so I will have it check: are there any holes that I'm missing? How do you think I could reword this better? Also use it with a lot of spreadsheets in terms of helping to organize some data that I may have, make more sense of it. I've used it for strategic planning as well.

Also as a thought partner — that's more and more how I'm using AI. There is a great book out there, Jeff Woods, "The AI-Driven Leader," and in there he talks about the CRIT model and using AI with this acronym CRIT: Context, Role, Interview, Task. You give AI the context in terms of what it is that you're looking for it to do. You assign AI a role — for example, I may tell it, "You are the director of product." And then you have it interview you: "Hey, I want you to ask three questions to really drill down on such and such an area to help me." And then the task — you'll tell it exactly what it is that you want it to accomplish. Using it as a thought leader, and that structure, has really been helpful. I've encouraged others on my team to do it as well.

Yes, there are a couple of things that are certainly keeping me nervous. One I've been reading a lot lately about is something called AI prompt injection. There are individuals out there that maybe are injecting hidden messages, if you will, or code into some of these models, maybe telling it to bypass certain instructions or behave a certain way other than what it was intended to do. Part of what scares me is, with companies who have adopted their own internal model that they're using, if there maybe would be an employee that would put some type of message in there and then things would go awry. There just needs to be constant monitoring, feedback, reviewing the models, the prompts, how things are generated, to help with that.

I also think that there could be some cognitive complacency. For example, if psychologists are confident and getting used to AI generating reports for them, and the outputs are looking amazing and they know that, "Oh, these sections are great, and it's interpreting the information I want, giving me the intervention I want," because they are so overloaded with work, they could potentially take things at face value and think, "Oh, I don't have to look at this closely anymore. I can do these 10 other things over here." That's part of the danger — not only with the psychological profession but with any profession. When you have AI used to doing something for you, we always need to go back and review: how can we improve it?

Yes.

I am most excited about the unknown and the potential — just the future in general. I think there are so many possibilities of what AI can do that we just aren't even aware of, and thinking about the potential jobs that aren't here today that could be created. There's a lot of opportunity, a lot of growth in the area. When you think about when social media just came on the forefront many years ago, people didn't even know what that was, and now all of the jobs that have evolved from that. I just think there's great opportunity, great excitement. I also think that we'll be able to scale products at a higher level, at a faster pace. I'm really just excited about all of the innovation. And in my world of psychological assessment, I'm really excited to see where we're going to go digitally. We've moved from paper to digital, and now with AI speeding things up, making things faster and improving products. I'm just excited to see where it goes.

Oh, sure. I think it is very anxiety-producing. Just thinking of work at companies — people are all over the place in terms of their comfort level with AI, their exposure, their use, their knowledge. There is certainly a lot of fear. And I do think though, if people don't embrace it, or at least explore it and educate themselves some way, no matter what profession they're in, I feel they're going to be left behind at some point, because there are so many other people right now that are taking this on, that are learning, that are educating themselves — and those are being the creators. Even if you think of psychological products that are being developed and other systems, if there's not, for example, a psychologist at the table, I think those products are going to fail, because there's that subject matter expertise, there are the ethical guidelines, there are all those years of training and schooling that the psychologist went through, where their input is really essential in creating these products.

Definitely team building for sure. Since it's a new tool, we're taking time to explore it, and it is helping us, I think, build that team-building environment. For example, we're taking a step back now and we're actually carving out space where we're dedicating an entire day to talk together as a team and just really dive in and explore and do a project, if you will, on anything related to AI. We are on a weekly basis having a check-in and talking a little bit about what are some of our successes, what are some of our failures. By doing that, we are creating a safe environment to really explore. I think it's building the teams up. We don't have these expectations right now like, "Hey, you just got to go out there and use this thing and do it," without any guardrails, without any guidance. So we're just really spending a lot of time trying to create an environment of psychological safety.

And then in terms of hiring, I would just recommend using it very, very cautiously. There is a person behind every CV, every resume that comes to you, that can't be captured. A lot of what people may be generating now and relying on AI is, "Okay, draft this CV for me, put this together, make this sound really great." That may or may not then really capture who that person is. So I think it is important to go beyond that paper and screen out, and try to find other ways to get to know people. Because in the end, it is really the human who is going to be doing the job.