Injury & Violence Prevention INdepth
Injury & Violence Prevention INdepth
Exploring AI's Role in IVP
In this episode, host Mighty Fine talks with Tracy Mehan, Director of Research Translation and Communication, about artificial intelligence (AI) and how individuals use it every day, whether we realize it or not, in our work and lives. She provides insights and examples of AI use in injury and violence prevention (IVP) work while also giving some tips and recommendations on what to watch out for while using it to ensure quality in our work.
Welcome to the IVP INdepth podcast, a Safe States Alliance production. I'm your host, Mighty Fine, and I'm thrilled to have you here. This is a space where injury and violence prevention professionals come together to share ideas, spark conversation, and dive into some of the most pressing topics shaping the field. So whether you're here to stay in the loop, find inspiration, or be part of a community that's seeking to make a difference, you are in the right place. And it's 2025 so I want to wish everyone listening a happy new year as we step into this fresh chapter. Let's embrace the boundless potential that lies ahead of us with new ideas to explore and meaningful connections to build. I'm super excited about today's episode, and I sort of feel like I say that all the time. However, I really am excited because we are stepping into the fascinating world of artificial intelligence, or AI, as most of us know it as. And keep in mind, this isn't a deep dive into all of the complexities of this topic, but rather a starting point, a chance to spark curiosity and get us thinking about how AI is shaping our lives currently and what lies ahead for the future. And joining me today for this conversation is Tracy Mehan. She's the Director of Research Translation and Communication at the Center for Injury Research and Policy and Nationwide Children's Hospital. So let's get started. Tracy, thanks again for being a guest on the podcast. So excited to have you. This is a topic that I'm super curious about, I find myself sort of exploring different articles, to read, watching things on YouTube, just trying to ensure that I'm staying as up to date on the topic as best as possible, because it seems ever changing. And of course, that topic is AI coming off of the holidays. Certainly, it dominated a lot of conversations that I had with my family and friends. So I'm glad that we have someone with expertise to talk to us about it today. So for starters, I think it'd be great for our viewers to hear from your perspective, what, what exactly is AI?
Tracy Mehan:Great question, and AI really is everywhere. When I talk about AI, people go straight to what they know most chat, GBT and the Large Language Models, right? And that's what we're hearing a lot about. But AI is bigger than that. It is in everything. It is in the algorithms that Netflix uses to suggest what shows we want to watch. It's in Grammarly, if you use it to edit text on your Word documents, it's in even on your phone, when you're texting somebody, it will suggest things, right? So AI is everywhere in our world, and some people who tell me they're skeptical of AI. They're not using it. You are it's it's being used on you, whether you realize it or not, right?
Mighty Fine:Unbeknownst to you, you're tapping it right, literally, right?
Tracy Mehan:I think, in terms of our conversation today, the types of AI that we're really thinking about, though, more will structure around those generative models that help you create new content based on information from their data set that they've learned from things that will can help you do translations, make your work go quicker, better,faster, those types of things,
Mighty Fine:I love that, right? It's the age old mantra, work smarter, not harder. Hard work is important, right? But we want to use tools like AI to help us work smarter. So thinking of that, how do people in injury and violence prevention in the field? How do they perceive AI, you talked a little bit about hesitations, but let's talk about the field specifically.
Tracy Mehan:Yeah, you know. So I talk in this field a lot, and I talk to people all around the United States, all around the world about this topic, and it really kind of falls into a couple of buckets. You know, there are people like us who are curious and want to know and are natural learners, but there's a lot of caution, a lot of fear and a lot of skepticism, okay? And I think a lot of that stems from a misunderstanding, or people just don't understand what these programs are and what they can do and what they can't do. So I think, you know, we're in this space where people aren't sure, and what I'm hoping this conversation will do is get people thinking about it, being a little more open to at least understanding what the programs are and how to think abou them.
Mighty Fine:Absolutely. So is there that there's that hesitation people are unsure, and often say, sometimes when we don't know, and this is the proverbial you when we don't know a we, I should say. Right is we sometimes fill in the gap with like, worst case scenario, right? Because you're just not well informed on all the possibilities and also some of the shortcomings. So thinking about that, what in your perspective, are some things that are possible for AI in injury, violence prevention, how can we use it to our benefit.
Tracy Mehan:Yeah, so I think the best way for people to think about it is kind of think about instead of being afraid that AI is going to steal your job or is going to put out content that is bad or negative, right, think flip that script and say, If I had more time in my day, what would I do with that, and let me use AI to help me have more time in my day or help my content be better, right? And so if you frame it in that context, some of the things that I've seen people in the injury prevention space use AI for are things like writing, yes, writing not to generate the content from scratch, right? I never do that. I write my own content, and then I ask AI to help me make it better, right? And so that can be, make it more understandable to a lay audience. That can be make it more professional, that can even be upload the contents of a grant and make sure that it's matching the grant requirements. You can ask it to do things like that, right? So if I'm using AI as the tool to help improve what I've already done, I'm still me. I'm still using my content that draws on my expertise, but I'm asking it to tweak it and just improve it. Right? Writing is one piece. Another thing is the kind of brainstorming aspect. So I use AI every day. I have it open all the time, and I almost use it as like this extra brain for me. So if I'm doing something like I need to do, create a campaign on drowning prevention, right? I go to chat, G, P, T, and I ask it, okay, I have to create this campaign, and I want to make sure I'm making this accessible to everybody. What are some of the health disparities that exist in drowning prevention that I need to think about, and it pulls out a list, right? It's not going to be perfect, it's not going to be all inclusive, but every time I do it, it includes information that I didn't think about necessarily, right? Even though I've been in the field for 25 years at this point, there's always something that it pulls up that I thought, oh, yeah, I didn't think about that. I gotta. I gotta think about that. Sure. And then the other thing around you know, AI that I've been using a lot for is accessibility. So for me, accessibility and my messaging is always really important, and that can mean a lot of different things. It can mean creating alt text to go with an image. It can mean what images I include. Right when we an injury prevention want to include an image, we're often stuck with what's in some of the stock photo sites and the stock photo sites are very limited in correct injury prevention behavior. They're they're very high end images. So you look like you live in mansions, and often you're white, right? Yes. And so what I've been starting to do is use it to generate images that allow me to include all different kinds of people in all different kinds of environments. And it's been really cool. What I've been able to do with it,
Mighty Fine:See, I love that the way, the more and more you talk, the more and more I think of it really, as a compliment or a support, not in the same vein, but I'm thinking of it like you've mentioned Grammarly or spell check. And no one's like, oh my gosh, don't spell check my work, you know. So people start to think of it as a complimentary tool to help support your work. And as you mentioned, part of what I've heard, and if you can speak to this, and just some circles I've been in, folks think of it as like a cheating right? And they're conflating that and and I think once we started think of it as a complimentary tool, just like we use other tools and tactics to generate ideas or to fine tune the work that we're doing, I think the better off will be. And I think the example you gave is quite perfect. It's like we see these images out there and they don't match what we're trying to convey, particularly around prevention, as you mentioned, certainly around diversity, across all aspects of diversity. So I think that's a great way to help folks to think about how we can actually use this to improve some of the challenges, or to address some of the challenges that we're faced with in our day to day work operations.
Tracy Mehan:Yeah, I think you kind of hit the nail on the head there. I think a lot of what happens is the first time we heard about these ChatGPT and llms was when. ChatGPT came out, and people started cheating on exams, right? That's how we heard about it, so that's how it got framed for us, sure. And I think it's been a challenge to kind of shift our thinking, but once you do that, and you start to think about the possibilities, I'm really excited about what it can do for us in the injury prevention space in particular. But you know, even in general, there's, there's lots of opportunities. If you open your mind and start to learn about what it can do versus what you shouldn't do with it. There's lots of options.
Mighty Fine:No, for sure, and if there's other things you want to elaborate on, but you talk, you gave her a lot of great examples. I love the one around, because we all know in this field, we're often writing grants and responding to grants and all types of requests for something, and it, it's a labor of love, let's say. And so if you have a tool to help support in that, in that courtship, so to speak, it's great, because, as you mentioned, most often, grants or other funding opportunities will tell you, Hey, these are the five things that we're looking for. And I think it's a great use of AI to once you've written the grant, help you to fine tune it, and also to make sure that there are there's congruence between what the application is asking for and what you have actually provided. So that's a great example of how we can be using it in IVP. And you gave some others. Anything else you can think of, of ways that we can use this to the best of our to benefit our work in IVP?
Tracy Mehan:Yeah, you know, another way I'd like to think about it is, what can I do to use it to save time, right? So, I mean, I have used it to help me draft let us letters to policy makers, to letters of recommendation, like even emails, even doing things like transcribing meeting notes, right? Yes, it's all those things that take time that I don't have to do anymore, right? So, you know, we all know if I have a draft of a letter, even if it's only 75% of the way there, it takes me way less time to edit that than it does to start from scratch, right? And and how good of the output you get depends on the prompts, right? So some of it is learning. How do you use this? How do you learn how to ask it a question in a way that will give you the information? There's all kinds of tips and tricks out there. You can google or follow different people on social media, and it's even things like, when you ask it a question, don't just say, write me a letter for this. Say,
Mighty Fine:like, I need it needs a little more than that, right, right? You can tell it
Tracy Mehan:I am in, you know, injury and violence professional, and I am writing a letter to this manufacturer, and this is what I want them to do, help me craft a letter, right? The more details you give it, the better. The response is yes. But some of it is just trying it.
Mighty Fine:Trial and error, yeah, and building up the comfort to actually do it. To that point, I recently took a course to help me to understand the importance of because I had never used it before. Interestingly enough, the first time I learned about ChatGPT in particular, it was over dinner with some friends, and he was jokingly, and folks don't judge me here. This is just a story that I'm telling you of how I got introduced to it. So he he wanted it to write a text message to to break up with his girlfriend. And it wrote it right. But then he was and he was showing us the whole progress, and I was blown away. Then he said, write it like a poet, you know. And so it made it more like, like a poet. Then he's like, like, write it how Drake would wrap it. He was just trying to show us, like, different iterations of it. And it was so cool. But it's, it sort of made me think about your point where it's about the prompts and understanding the importance of prompt engineering, I think I've heard it called or something like that, and recognizing similar to what we learn in Epi garbage, in garbage out, if you give it like a terrible prompt, then you're not going to get the best output. But if you really think about those prompts and and sort of trial and error, probably to some extent, then you'll end up with a better product at the end.
Tracy Mehan:Yeah, for sure, practicing being open to learning about it, having conversations within your team and your organization, I think, are vital.
Mighty Fine:So let's talk about that for a second. Tracy, thinking about those conversations to your point, I think the way it was fed to a lot of us is like, hey, it's a tool. It's for cheaters. It's for people who are unintelligent, who don't know anything, and this is going to do everything for you. So I'm smart, and I have my PhD in neuroscience, and I wouldn't dare use chat sheets, you know, just being facetious here. But in all seriousness, help us. To think through our listeners ways that they can broach this topic with leadership or colleagues, or I found too sometimes when people use it, they're afraid to say they actually use ChatGPT, because people are still sort of on the fence on the benefits of it. So if you can talk to like, how we can even start those conversations in the workplace, that would be helpful.
Tracy Mehan:Yeah, you know, I'm going to start with a story. Okay, you did that as well.
Mighty Fine:Yes, we love stories.
Tracy Mehan:You know, this is an area that I've been really interested in. I've been talking to people about it. I talked to our own team at my institution, right? And we have summer students in every summer, and usually I give a presentation to them where I'm talking to them about health communication and how to talk about their research. And, you know, while ago I started talking about social media. Well, this year, for the first time, I was going to talk about AI. So I wanted to talk to them about what does this look like in injury prevention and research. And I let all of our staff know, like, Hey, I'm going to be doing this. Our students might be asking you questions about it. And I heard, well, they're not using AI when they're working with us, so it doesn't really matter. And I said, you know, are you sure about that? Because, have you asked them? Because I bet they are. And they said, No, same thing. You said, that's used for people who are trying to who can't write, who are cheating or, you know, doing things different than that. I said, No, I think it's interesting. You know, let's have the conversation. Ask your students, and they came back and said, huh, our students are using it. We didn't know, and to me, that's really important, right? Because even some of the journals that we're starting to submit to now, they ask us the question, was aI used, and if we haven't had a conversation with our collaborators, with our students, with other people we're working with about whether or not they've used it, are we accurately answering those questions, right? But secondly, to have the conversation about whether or not it's okay and when it's okay, right? Yes, we talked about Grammarly. That's actually AI. Yeah, so did if you use it to just proofread, are you using AI and is that okay, right? What does that look like another we talked about images, right? If I create an image that I want to put into a journal article or publish with a blog or something like that. Am I allowed to do that at my institution? There's some confusion on that. Are we allowed to do it? And if we do, do we have to disclose that it was created by AI, right? Yeah, and this is going to be more and more of an issue. Some of the stock photo sites, like Adobe Stock are using AI generated photos and selling them, and if you're not careful, you might not realize it, right? Yes, I had that happen recently. We were looking for a photo for Child Passenger Safety, and it's really hard to find correct photos. And they found one our blog editor, and it was aI generated, and none of the docs that reviewed it realized it, and I was the only one that said, Hey, this is AI generated I actually think it's great, and I think we should use it, yeah, because the ones that were quote, real all had an error in it, and the AI generated one didn't right,
Mighty Fine:See it speaks again, to the benefits, I will say, on social media. I've seen some AI generated videos not to get go down that rabbit hole. And some of the people like, Oh, this is so great. I'm like, do you know this is not real, right? But to your point, I think it's worth sort of uplifting some of the challenges that people have or have seen, or some of the things that we want to avoid so that at jobs, we can craft policies or procedures of how to integrate AI into our work in a healthy and meaningful manner, addressing some of the misgivings that folks have, or trying to avoid some of the aspects of AR that are less favorable. But I to your point, though, I think in this space that we're talking about, one of the things I see as challenging is having a ChatGPT, or some other prompt engineering system, generate something for you from scratch, and then passing it off as, like, you know, original works, which I'm not seeing as much, but I know that's a concern that I've heard from folks, particularly when it comes around, like writing for a journal, and there was a round table that I was on, and they told me, at their last editors meeting, they talked about, and this seemed, well, I don't want to say it's extreme, but in my mind, it's extreme, but maybe not for others, is having the AI tool listed as one of the authors on the paper. You know, which I thought was an interesting concept to think of, but it was a discussion that they were having, having among editors of a particular journal.
Tracy Mehan:Yeah, I think we're in such an interesting space right now. It's people don't know what this is. Don't. Know how to use it. Don't know when to disclose, when not to disclose, what that looks like, right? So to me, I think just having the conversations and being more open and talking about it, right? So on my team, the way we're doing it is, whenever I have a new staff member or an intern or a new collaborator, I upfront have the conversation. This is how I feel about AI. This is how I use it. This is what I'm planning to use it for. Yes, share with me, right? Yeah. And then in our team meetings, we talk about, every time we see something new with AI, we share, Hey, I saw this cool thing that we can do. Have you thought about doing that? Let's brainstorm, is this good? Is this bad? What do we need to think about around using it in that way, right?
Mighty Fine:I love that because it normalizes it and it allows people to feel safe. I'm using air quotes here, folks and using it and that their their intelligence or their intellect or whatever, won't be challenged because they decided to work strategically, which is wild to me when I think about it, but I also recall from the training that I took, and you can speak to this better than I'm certain is my trainer told me, the more and more you use the tool and you give it prompts, it starts to even learn your writing style and what you prefer and what you don't like. So you find yourself as you're building and curating content you're like, This really sounds like me. You know, because you've you've fed it the prompts it, you've fed it your writing and your writing style, and you've prompted it around that. So I don't know if there's anything you can speak to around, sort of the importance of training and using it, but also how the continual use of it even just sharpens or develops your own skill and your connection may not be the right way, but your connection to the tool, you know.
Tracy Mehan:Hey, I'm very polite every time I talk to my ChatGPT, just because you know you never know right? I do, though, feel like it does start to know you right when I ask it questions now it will give me answers in a way that are helpful to me. It does help my writing have my voice a little bit more. One potential downside I have noticed, though, is, you know, injury prevention can be a narrow field sometimes, and I have been using it to try and brainstorm, who can I talk to in the field about, you know, this topic, and it's has started to narrow it down to people that knows that I have worked with before, and I actually don't want that, so I have to tell it, no, you have to broaden the spectrum, right?
Mighty Fine:These people, other people,
Tracy Mehan:Right? So there are some cautions with that. And sure, we've been really positive, mighty. But there are some things I want to make sure that we are thinking about too in terms of please, please these discussions, right? I think it is important when you use these to be aware of some of the limitations. So one example is, I used it to create some messaging on smoke detectors and smoke alarms. And I wanted to see just what will it give me, and what are the recommendations, and is it following best practice? And it sounded really good, and it was really close, but it didn't actually follow what best practice recommendations are in terms of where you put smoke alarms in your home, right? Got it. So you have to be really careful, because it sounded really good. And if I didn't know, I would have thought, huh, that sounds great. I should do that, right? So you have to be really careful about making sure that you're really looking at what best practices are in injury prevention, because we know that can make a difference, right?
Mighty Fine:Yeah, that's a great point. It's, I don't know if it's the same, but in the course, I learned about how sometimes it can hallucinate and just make stuff up. That sounds accurate, but that that was one of the the pitfalls that he encouraged me to ensure that I avoid, and it's he also mentioned that you shouldn't use whatever the final output is from, say, like ChatGPT, that shouldn't be your final submission. There should always be, like, another cross check to address some of the things that you were saying. But sorry, you were going to talk about some other things related to that as well.
Tracy Mehan:No, that's great. I'm going to give one more example with the with hallucinations. Many people don't know that the free version of ChatGPT, at least currently, the training only goes up to September 2023 so anything that is before that is great. If you're asking after that, it will give its best approximation, but it isn't actually based on data and where that matters. An example that I use is, if you're asking it, for instance, what are the number of firearm related injuries in the United States? It will tell you a confident answer, but it won't be accurate, because it's guesstimating based on what it knows from before, right? And if you don't know that, that's how it works. You. Might say that that's accurate if you don't do your due diligence and data check. Now it's getting better. The paid version can have more access to the internet. You know, that's even a little sketchy. It now has a little if you're in the paid version. There's a little globe at the bottom that you can click on where you can tell it to go out and search the internet, which is kind of a newer feature. All of those are good, but what I always say is go back to the sources and figure out where the data is from before you use anything. So I think that's really important.
Mighty Fine:Yeah, and I think that's a great point to drive home is it's that it's while it's generating great content and building upon what you shared it, it's always important to not send that off. A friend, we were talking about it, and she told us a bit of a joke where her manager generates in ChatGPT. You know how it has the Asterix. He would send everything and forget to take out the right but it goes, but it's sort of adding to our point that there should be another review prior to finalizing whatever you've asked it to generate.
Tracy Mehan:Yes, I tend to look at it more as like an intern, like I write my own stuff and I have it polish, versus having it right for me, for you, or I'll use it to brainstorm, or we talked about generating images while we're talking about the images. One more thing I do want to really make sure we talk about is just thinking about the the bias in these programs. For sure, I've had some interesting conversations with people who say to me, Well, they're not biased, because it's computer generation generated right if I'm in front of you and you see that I'm white or I'm black or I'm a woman or a man, right? We have our preconceived notion, and they say the computer doesn't do that, but what they're not thinking about is the how the information is trained, right? These models are trained by humans, so what we feed into the programs to train it is bias based on what we include and what we don't include, for sure, their bias based on the trainers. Right? All these programs have trainers that make them better, and yes, each trainer comes with their own biases, right?
Mighty Fine:There's a still a human element to it. For sure.
Tracy Mehan:Still a human element to it. And even in my own explorations, when I ask it to create an image without giving it any other direction, most often it will be an image of a white person. So you have to actively choose to ask for something different. And that's part of what I do all the time. And you know, we can use it to combat and have more images with more representation if we're actively choosing to do that, yes, but we need to be aware there are biases in the language, in the images, and everything it puts out too. So just acknowledging that, being aware of it and knowing how to counteract it can help make it much more useful.
Mighty Fine:Awesome. Awesome. That's a great point that I don't think is talked about enough. So thank you for elevating that, and it's been great chatting with you. We talked about the benefits, some of the challenges, how to start the conversation, and really how to sort of remove the stigma around ChatGPT, and actually recognizing, yes, there are challenges, but look at all these benefits as well. And if we know the challenges and our use of it, we can work to avoid or prevent those so I'll stop there, but give you the last word. Is there anything else regarding AI that you would like to tell your fellow colleagues in the IVP space?
Tracy Mehan:Yeah, I would just like to say that people are using these programs, whether we are or not, so even if you don't want to embrace the use of the program, I think it's really important that we understand what they are and how to use them, so that we can talk to our communities, our patients, our people that we work with about how to use them and what you can and can't do with them, right? But I also hope that as an injury prevention field, sometimes we are a little behind the acceptance curve in being forward thinking in our innovation, right? And it is here and it is being used, and I think we can be on the forefront of some really cool things if we embrace it, start to think about how we can use it and what it can do for us.
Mighty Fine:Yes, I love that. I think that's a perfect way to close - that it's coming. Let's embrace it and see how we can even get ahead of it, to show to use injury and violence prevention as an example of how best to use these tools and strategies. So thank you again, Tracy, it's been a pleasure. I could talk to you forever about this, so we may have to ask you to come back for part two, and I'm certain that the listeners got a lot out of the conversation as well. So thanks again.
Tracy Mehan:Thank you for having me. I'll come back anytime. Awesome.
Mighty Fine:That's a wrap for this episode of IVP INdepth, and thanks for listening. A big thank you to our sponsor. The Society for the advancement of Injury and Violence Research, or SAVIR, which is a professional organization that provides leadership and fosters excellence in the science of Violence and Injury prevention and care. We want to thank them for supporting this conversation and their commitment to injury and violence prevention overall. Learn more about the fascinating work that they're doing by checking out their website at the savir.org that's T, H, E, S, A, V, I, R.org, and if you're not already a member of Safe States, now is the perfect time to join be part of a community working to create safer, healthier communities for everyone. You can check out our website to learn about membership, but also explore the range of resources, including toolkits, training opportunities and other useful information designed to support your work in the field. Please don't forget to hit that subscribe button so you never miss an episode, whether on Google podcast Apple or Spotify. And while you're at it, leave us a review to let us know how we're doing as always, thanks for tuning in, and we'll see you next time. Until then, stay safe and injury free.