Behind the ABA Talk Ep. 7 - Efficiency Meets Ethics: How BCBA's Can Responsibly Leverage AI Tools with Guest Jennifer Magnuson

Join Lisa Gurdin, founder of LSGurdin Consulting, for heart-to-heart conversations with expert speakers from the LSGurdin Consulting Continuing Education program. In this episode, Jennifer Magnuson, BCBA, shares her journey in ABA and provides practical insights on how behavior analysts can responsibly integrate emerging technologies into their practice. 

With over 20 years of experience supporting individuals with autism and developmental disabilities, Jennifer has spent the last two years intentionally designing programs that support young adults transitioning out of high school and into adulthood. Her approach focuses on partnership, goal achievement, and creating a higher quality of life, while also building inclusive networks and communities where people with autism can contribute and thrive.

Register now for Jennifer Magnuson's CE event: Efficiency Meets Ethics: How BCBAs Can Responsibly Leverage AI Tools onApril 3, 12-1:30 PM EST. This event will also be available for on-demand CEUs. 

Register at https://behaviorlive.com/organizations/lsgurdin/LiveEvents 

Connect with Jennifer: 

https://www.linkedin.com/in/jennifer-magnuson-5529a6176/

https://www.runthegamut.org

Follow, Connect, Reach Out:
https://www.linkedin.com/in/jennifer-r-9a327434/
www.linkedin.com/in/lsgurdinconsulting
https://www.tiktok.com/@lsgurdin_consulting
https://www.instagram.com/lsgurdinconsulting/

Transcript

SPEAKER_00 0:14

Welcome to Behind the ABA Talk Podcast, where we have heart-to-heart conversations with ABA experts, thought leaders, and change makers. Each episode unpacks the stories, insights, and behind-the-scenes realities that shaped our experts' professional journeys. I'm your host, Lisa Gerden, founder of Alice Gerden Consulting, and I have a question for all of us to consider. Do you want to dive into the world of AI to do your work faster and with less response effort, but you don't know where to start and you are worried about falling into ethical pitfalls? I don't know about you, but my answer is definitely yes. Fortunately for us, we have Jennifer Magnuson here to share her recommendations on how behavior analysts can responsibly integrate emerging technologies into their practice. And I think the conversation will be relevant for professionals and across disciplines. Jen will be presenting this topic on April 3rd as part of the Alice Gerdon Consulting CE program. She's a board-certified behavior analyst with over 20 years' experience supporting individuals with autism and developmental disabilities. And she currently serves as chief of staff at Run the Gamut, a nonprofit that supports young adults with autism and their transition to adulthood. Jen, I am so happy you're here today.

SPEAKER_01 1:39

I'm happy to be here. Thank you for having me.

SPEAKER_00 1:42

Yes, absolutely. So if you could share your journey in ABA, that would be wonderful.

SPEAKER_01 1:50

Let's start there. Great. Well, it started with a puppy, which I think is fun. When I was a junior in high school, I, you know, did the thing where you go to your guidance counselor and they have you take a quiz to tell you what you should be when you grow up. And she, my quiz told me that I should be a cable truck repair man person. That's a new one. You know, I don't know what I answered on the quiz that gave me that result. And I often wonder what would have happened if I went that direction, but I didn't. But I came home and that same day we had gotten a puppy. We was like the day that we picked him up. He was this like adorable little black lapid or retriever, um, just like this chunky little, like, how could you not fall in love with it? Um, and the person who we adopted the dog from gave us a calendar that was um, I think it was like a fundraiser from an occupational therapy association that each month was a picture of a lab, um, but they were supporting in various capacities people with disabilities. So opening drawers or sidewalk assistance or all the different ways that service animals can support people. Um, and so I thought I was like, well, my guidance counselor wants me to try the cable repair truck, um, but this looks like a lot more fun. Um, and so kind of that moment, just being like this cute puppy and realizing that that could be a tool that helps people really kind of started me on this career path. So I went to school for psychology, um, did kind of all of those things. And then when I was graduating from my undergrad program, um I was really kind of like, well, what's next? Um, and the the choice was to either move back home, which I didn't necessarily want to do, or go to grad school. Um, and so I was very fortunate to find at the time that the May Institute was sponsoring a program with Northeastern University. So I was able to work um and get really good kind of supervised field work experience while also taking classes at Northeastern. Um so I did that program, not really knowing a whole lot about what ABA was, I just kind of jumped into it, um, but really fell in love with the field. Uh, the students that I had there, um, the way that I was learning in class, really practical, tangible, hands-on things that I literally was taking into my classroom the next day and seeing student success with, um, really just I think solidified that this was the field that I wanted to be in and working in. Um, so I started there. I worked in a lot of different settings. I really wanted to um, you know, I jumped into the field really young and really early. So I wanted to kind of explore different settings. So I actually went and worked at a hospital. Um, I worked in a couple of different school settings and in some different agencies. I did a little bit of everything to get to where I am today, which I really kind of came full circle and reconnected about two years ago with one of my friends from grad school. Um, and she was at the very beginning of building this new transition program. Um, and so I partnered up with her, and we've spent the last two years really thinking about and intentionally designing a program that supports young adults as they're transitioning out of their high school programs and into adulthood. Um, and so we're really thinking about how we can use principles of behavior analysis to create this environment that really supports that transition from high school programs to really the ultimate goal is employment and you know, independent living with the supports that our students need. Um, so I'm really excited to be where I am today. And I feel like a lot of what I did early on has really kind of has built to this point right now.

SPEAKER_00 5:41

That's amazing. Um, I think we we almost overlapped at Northeastern back at the um we were so close in the MAVA program. And I remember being in grad school also like with light bulbs, like kept going off in my head, like, oh, that's what that is, ooh, that's what that's called, right? Like, and then using what we just learned. Um, I had that same experience and just really falling in love with the with the field as well. So um, so when you think about where you are today in in adult services, which is amazing, and and I've seen what you guys have built, um, and it really is an impressive place and an impressive program. What for you does this uh world of adult services, which is you know its own thing, different from you know, school-based services, um, what do you love about it? What is reinforcing for you? And you know, what's your why of what you're doing right now?

SPEAKER_01 6:44

Oof, that's a big question. Um I think you know, my first answer is the students. It is, I'm calling them students, they're young adults. Um they're so cool. They're just, you know, they're fully formed people by the time they leave school and they know what they want and they definitely know what they don't want. Um, and being able to think about work with them in a way that's partnership and not um, you know, like a traditional teacher-student model, but really just thinking about like you're a person with a goal, and I'm trying to get to support you in reaching that goal, um, I think is really exciting. And it's something that um I think working with adults in particular, you're really making some meaningful steps for people and really connecting them to a higher quality of life, which I think is kind of the center of all the work that we're doing. Um, I think my why has expanded though too now. It's it's less about just kind of helping individual people meet individual goals. And it's more about kind of creating a network and a community that can really kind of just embrace inclusion and really embrace this idea that people with autism have so much to contribute to the workforce, to the community, um, and kind of just building out a world where that's that's a like commonly known thing. Um, and that feels like really meaningful work that we're doing at Run the Gamut right now. Um, we've been going to events at the State House, we've been kind of partnering with different agencies and different kinds of advocacy groups, and it's really exciting to see the pockets of work that are happening everywhere. And I really hope and I feel like part of what we're trying to do is bring all those communities together. And so it doesn't feel like little pockets in isolation, but it feels like a whole network of people doing really good work together with kind of shared and common missions. So it's, I think I have like a new why um in this kind of phase of my career, just because it's been so exciting to see all that work come together.

SPEAKER_00 8:51

Absolutely. And I there's so much value to uh creating partnerships on so many levels. I mean, you mentioned the partnership with the with the students, with the young adults. Um, you have it run the gamut. I think there's partnerships with families, partnerships with community and advocacy organizations and other people doing amazing things. And it's so great to see not only see all that happening, but really be a part of that and a force to help move that forward. So it's awesome. I love it. I love it.

SPEAKER_01 9:29

Yeah, definitely.

SPEAKER_00 9:31

So seeing where you started with this idea, uh, you know, the puppy kind of sparked your your your journey and where you are today, what would you tell your younger behavior analyst self?

SPEAKER_01 9:45

Oh um slow down. Yeah right. Like I think just take the time to really understand before jumping into action. I think you know, early on it was here's a problem, here's an intervention, do it quick, right? Like fix the problem, um, you know, kind of just react to what's happening and you know, you have all these tools, so use them. Yes. Uh and I think, you know, I'm grateful for the direction the field has moved in terms of kind of cultural competency and compassionate care and really just think about the context that the person is in. Yes, yes, the impact that your work might have not only on that person, but the other people kind of in their circle, and make sure that what you're you're recommending or what you're implementing or what you're doing is really in line with what that context is. I think, you know, it's this is not a one-size-fits-all world, and so our work can't be either. And so I think I would just go back and tell myself, let's just slow it down and make sure we really are fully understanding before we're jumping into action.

SPEAKER_00 10:59

I love that. That is so important, and it's often a place we miss, a piece we miss because everyone around us and ourselves want to fix the problem, right? And they look to us to fix that problem. And it isn't a problem that we should react to, we should respond with thoughtfulness and taking that pause is so valuable, absolutely. Um, and and so on April 3rd, you're gonna do a CE, and you are going to uh talk to us about specific strategies for how we can use AI with the goal of using it for efficiency, right? Um how we can do that while maintaining our professional integrity and person-centered values and all of the things, um, and not get kind of sucked into the efficiency of it while forgetting the ethical impact and the ethical considerations, right? Because I think it's easy, it's so easy to use and to get sucked into that to that space. Um, and so I'm I'm so excited to talk about that. I'm wondering how AI um has become kind of a focus in your work.

SPEAKER_01 12:17

Um yeah, yeah. Um, well, to tell the story, I have to start by admitting an embarrassing mistake that I made. Um, I was using AI. Um, and let's see, it was probably about three years ago. Um, I was leading the clinical team at a at a human services agency with we had about 500 uh direct care staff working directly with our students or clients. And I was leading a clinical team and I wanted to develop some really meaningful, like mini micro trainings. Like, can you put together a 10-minute training on how to use an AAC device for modeling? Right. I wanted my SLPs to kind of develop that content and then go take it out and do it. Um, I asked my mental health counselors, can you develop, you know, 15-minute training modules on trauma-informed practices that you can do, like things in the moment that you can do and take it out there? Um, and so I kind of had an outline of all the different types of trainings that I wanted people to do. And I was sending off emails, right? I'm just multitasking, working quickly. I'm at my desk, sending out emails, you know, to my speech person. Can you put together this training to my senior VCBA? I wanted to, I wanted them to do a mini module on the dimensions of reinforcement, right? It's not just good enough to deliver reinforcement. You have to think about how and what and and how long from when the behavior happened to when you delivered the reinforcer, right? So I really quickly just wanted her to put together this training on the dimensions of reinforcement. I was newly experimenting with the use of AI, and I thought, you know, instead of crafting the perfect email, I just typed into chat really quick, can you write an email to my senior BCBA asking her to put together a 20-minute training on dimensions of reinforcement? Copy, paste, send. Literally two minutes later, she writes back, What the hell are you talking about? And I was like, What do you mean? You're my senior BCBA. How do you not know what the dimensions of reinforcement are? So I went back to my email and uh chat very confidently decided to list out the four dimensions of reinforcement within the body of this email that I sent without fully critically reading. Um, and they were completely wrong. They sounded great, but they were not the ones that we know and love and use. Um, and so I totally just had this moment where I was like, whoa, I just a hundred percent trusted this system that I literally know nothing about, and sent off an email with totally incorrect information. And then, you know, that moment has just stuck with me. It's it's so tempting to use it. Um, and it speaks so authoritatively and it almost always tells you yes. And so I realized in that moment that at first I wanted to just throw it out and be like, okay, I'm done with it. It's just, it's not gonna work. Um, and then I was also like, it's not going anywhere. So instead of just throwing it out, I realized I really needed to understand how it works in order to be able to use it. Um, and so I spent some time over the last few years really just learning kind of the background of what it is and what it means or how it gets to the answer that it gives you for things and understanding that gives you kind of an ability to use it as a tool. Um, and I think, you know, there's lots of like time-saving efficiency things that it can do for us, and there's a lot of concerns to be held with it. So I think it's an important conversation that we're seeing a lot. Um, and so I'm looking forward in you know, in April to be able to talk a little bit more about it and share some more of that experience with everybody in your in your CE group.

SPEAKER_00 16:01

Yes, absolutely. There is so much to talk about. What would be one way that you've used AI um successfully, you know, where you were like, yes, I got this, is this is what AI can do for me.

SPEAKER_01 16:19

Um I think I use it as an iterative tool, meaning I never accept the first thing that it gives back to me. Um, but I, you know, through feedback and kind of back and forth processes, you can get to the finished product that you like. Um, and so I think a lot of it is kind of giving it really good and well-crafted prompts. And that's some of what we can talk about on the third, um, to make sure that you're giving it the right information and that it's looking at the right thing, right? So that day when I sent off that email, well, I shouldn't have said draft an email on the dimensions of reinforcement. I should have said, pull up the peer-reviewed research article by Neef and Lutz from 2004 that outlines the four dimensions of reinforcement. Use those to write an email, right? So a lot of it is about giving it the right information or giving it the context that it needs to be able to do the job well. Um, and learning those and kind of learning kind of the four there's formulas for how you craft your prompts. And if you can do that and then think critically about the responses that you get, um, you'll be a lot more successful with that.

SPEAKER_00 17:31

So interesting. So interesting. Um so when we talk about the concerns, what are a few of the main concerns that you see with regards to ethical use of AI in in our practice?

SPEAKER_01 17:51

Yeah, I mean, confidentiality and privacy are like is the obvious one, right? Like don't put student names into chat. Just don't do that. Um, don't put any identifying information in there. Um, you know, I think that's one that people are are pretty aware of. Um, I think this idea of accuracy is really important to think about. Are you double checking? Are you, if it's giving you sources, are you actually clicking and looking at the sources it's giving you to make sure that one, those sources exist, and two, they say what you know, what it summarized it to say. So thinking about the reliability and really critically thinking about the information it gives you. Um I think one that's really tempting is to think about using it to expand your scope of competence. So I know early on and in a lot of kind of ethics conversations as behavior analysts, I think we do a really nice job of talking about operating within your scope of competence. Um if a case seems outside of it, find somebody, right? Refer it out, do your own professional learning, right? Like don't operate outside of your scope of competence. Um, I think chat is a really kind of tempting avenue to say this one feels a little bit out of reach for me, but what would you do? And it it tempts you to maybe expand a little bit beyond what your kind of current scope of competence is. So I think we need to be really careful about are we using it to support the work that we know how to do, or are we pushing boundaries a little bit with it? Because I think that's an area that I would I would want to recommend people be careful with and making sure that you're still seeking either that consult or that referral or um really credible professional learning and not just saying, like, as a BCBA, what would you do if, and then kind of incorporating that into your repertoire. So I want to be careful about that. Um, and then a last one that I think is important to think about is just the bias that is inherently in a lot of the large language models. So they're only as good as the data that they're trained on. And this is a little bit of what I'll talk about on April 3rd, but um, they they're not brains, they're not critical thinkers, they're data analysis, right? So they get data and then they spit out more data. Um, and so if you're asking it questions, it will give you your answers with whatever biases in the data set it was trained on. Um, one that I think about as a behavior analyst a lot is um there are a few research studies out there that have looked at um the results of functional analysis or functional assessment across kind of a larger population, right? So there's one study that looked at I think 160 some individuals or behavioral assessments, and they found that, you know, escape was the maintaining variable for I think like 30%, attention was like another 30%, but only 9 or 10%, I need to look up the article, only 9 or 10% of the behaviors were maintained by a tangible function, right? Which for that 160 population, that was the breakdown of function. If chat has that information in it, it is going to make an assumption that 30, 30, and 9% of cases line up with those functions. So if I plugged in my kid and my information, it's probably going to bias its response toward an escape or attention function and not that tangible function, just because I've told it or the training model has told it that it's more likely to be attention than tangible, which maybe it is, but for this individual, I don't want that bias in my interpretation of results. So I think if people are starting to use it to analyze data or to kind of help with clinical decision making, that's really dangerous territory, especially if we're thinking about bias that is introduced in that training set and then in the responses that it's giving us. So that's I think a big part of what we'll be talking about on the third is really being careful not to let. That bias influence clinical judgment.

SPEAKER_00 22:03

Yes, it's sort of like adding bias on top of bias, right? Because we have, we come to the table with biases. Everybody does, right? So then you're adding computer-based bias, um, technology-based bias on top of our own biases. And that's like a lot to filter through, I think. You know, it's just a it's a lot, which is what I think, you know, all of these things you're mentioning makes people very, very nervous. And I think you're also, you know, when you talk, I like what you said about um using AI to help us do what we know how to do better or use it in a different type of uh, you know, um, using it as a tool to do what we do, right? And not to try to do something different. And I I really, really appreciate that. That makes sense to me. Um, so what would be one way do you feel like we could we're not gonna put clinical information and use it to guide clinical decision making? What do we use it for? Like what's an example of what we would use it for?

SPEAKER_01 23:13

Oh, there's so many things. It's a lot of time saving, right? So think about in your clinical practice where are the places that you lose a lot of time, right? What are the things we're doing that um take the time away from the stuff you really like to do? Um, and honestly, it sounds simple and easy, but I think the material making is such a huge time set. Um, we have here at Run the Gamut, we have a lot of kind of experiential learning spaces. Um, and we have one room. I I know a lot of programs do this, but we have one room that's set up like a stock room. So, you know, I spent like two months saving up all my cereal boxes and yogurt containers and all that kind of stuff. And we've we've built this room, it looks like a store or a stock room. Um, and so one of the job simulations that our students do is order fulfillment, right? So when I do an online order at Target and I go pick it up and they bring it out to me, uh, somebody went through the store and shopped for all the things at my cart. And so I wanted to create um different order fulfillment slips for our students to take one, practice with, and go. Um, but I knew I would need a lot of them for built-in generalization, for to keep it varied so they don't just memorize the slip and grab the same thing every time. So I just made a list of all of the things in our stock room. And then I put it into chat and I said, I need 200 order slips that use um, you know, eight to 12 items each varied across this list. Um, some order slips should have an item I don't have that the student can make a substitute. Like I gave it rules about what I wanted. And in two minutes, I had 200 order slips that I literally could just copy, paste, and print out. And so that whole exercise took me what, five, 10 minutes when you know, 10 years I would have sat there in a Word document typing out yogurt, cereal or detergent. Next one, you know, like it's just there's so much time you can save. Um, some of the image generation technology or work is getting pretty good. So if you're writing a social story, you can say, generate an image that shows two eight-year-old boys on a playground arguing over a soccer ball. And it will do that for you. Um, again, you have to be good about your prompts to get what you want, but the time that I used to spend literally just searching Google images, trying to find like a palm tree with a white background or whatever, because I needed it for some kind of you know training program or something. Um, now it can just generate the image that you want. So uh time saving and material making, I don't think can be overlooked because of just how much time we spend on that and how specific you can be about it. Um, that's not as like fun and exciting as some of the other uses. Um, but I think it's one that a lot of might resonate with a lot of people.

SPEAKER_00 26:05

Oh my gosh, for sure. I mean, I my mind is being, you know, kind of blown away a little bit of like, oh yeah, that's a good idea. Oh yeah, you know, and that material making, or to me, I've been playing around with it with flow charts, you know, putting in a uh, you know, step by step of what's staff should be doing to support students and coming up with a visual, right? So that's something that AI can do, right? And the material making, I mean, for me to make a flowchart, it would take me a very long time. And I'd start over about 15 times because I'm not getting it the way I want it to look. Um, and what I want to say, what you know, how I want it to look. So I oh, I I just appreciate that so much. And I'm so excited um about hearing more on April 3rd. And you've given us uh a lot of teasers as we've talked, um little nuggets of what we can expect. Um, is there one other kind of primary message or primary piece of information we can look forward to?

SPEAKER_01 27:13

Yes. Um, so I did a thing for this that I'm a little bit excited about. I, you know, a big part of behavior analysis is being conceptually systematic, right? Building on the work that's been done before us. Um, but this is this is new. There, you know, I can't go back 30 years and see what people were doing. Um, but we do have, you know, five, 10, 15 years of research both within and outside of our field. Um, so I actually did a little lit review. Um, I spent an afternoon kind of combing through, and I I have about 400 articles that I kind of pulled from the literature that all have some kind of connection to AI or large language models because I wanted to see, you know, there's a lot of like me, people doing continuing ed and talks about AI in your practice. Um, but I wanted to go back to the peer-reviewed literature and see what's happening there to help us start to begin this idea of conceptually systematic application of AI into behavior analysis. And so I have some really nice takeaways just from that exercise that I'd like to share with the group to give us just that kind of strong foundation to start our conversation with.

SPEAKER_00 28:30

Yes, absolutely, absolutely. And I uh love the innovative way that you are taking our science and our approach and and and placing here in understanding uh AI, uh using our paradigms, uh, you know, and um so that we can use it like like I said, in a very ethical way. And and I think one of the things that's important as well is thinking about the human piece of it, right? So as we use AI, making sure we're also, you know, like you said, use it as a tool, and we are really focusing on the human piece of what we're doing and continuing that that that emphasis on helping people, you know, make those connect connections, find community, reach their goals, um, and just experience meaningfulness in in their life, whether they're kids or adults and absolutely, yeah.

SPEAKER_01 29:31

That's it. That's it, right? So um, if there's a tool that helps me do it better um and more efficiently, that's great. Um, and caution, Bernie, you know, we're needed.

SPEAKER_00 29:45

Yes, absolutely. And um, I'm excited once again to hear your talk and to be able to do things more efficiently to focus on, like you said, the things we love to do uh as practitioners. So um thank you so much for being here, Jen. I really appreciate it, and I've learned so much just within this uh conversation, uh, and we'll be adding to it on April 3rd. Uh and this has been Behind the ABA Talk Podcast with Lisa Gurden and our wonderful guest, Jennifer Magnuson. Thank you so much for giving us an ethical approach to AI so we can be more efficient in our work with all without all of the worries. And be sure to join us for Jen's upcoming Ellis Gerdon Consulting CE event on April 3rd from 12 to 1.30 Eastern. And if you miss it, no worries. You can always watch it and earn CEUs live or or on demand. You can go to www.lisgirdonconsulting.com to learn more about these meaningful continuing education events. Be sure to check out the On-Demand Library and our newly launched RBT modules for PDUs, and those modules are also intended for support staff who are working in schools and agencies and across all different settings. Thank you to our CE partners and sponsors, Massachusetts Partnership for Youth, Lab Collaborative and Accept Collaborative, and a special thank you to my podcast producer, Cole Asher Hoffman. See you all next time.