With exciting new advances in generative AI platforms like ChatGPT, there is a world of new opportunities for us to explore. However, ECPs (and professionals in all industries) have been left wondering how exactly to apply these technologies to their day-to-day operations.
Enter Keri Sculland. Keri is the Manager of Editing and Content Strategy at Marketing 4ECPs. In this episode, Keri dives into how her team has been using various forms of generative AI to create marketing content, as well as the current and future applications for ECPs.
Stay tuned for multiple other episodes on the topic of AI in Eyecare coming soon!
We will be interviewing optometrists, engineers, and scientists to discuss current and future AI-based tech that will change the way we provide eyecare.
Be sure to LIKE, REVIEW, and SHARE!
—
Watch the episode here
Listen to the podcast
Generative AI Applications In Eyecare With Keri Sculland
Thank you so much for taking the time to join me here to learn and grow. As always, I’m so grateful. I’m going to have a quick request for you right off the top. If you find value in this episode, which I’m sure you will because this is going to be a fun conversation, please share it. Please send a link to your friend, put it on LinkedIn, take a screenshot, and put it up on your Instagram story, something or other. Let people know that we’re having this conversation because there are a lot of people out there who are going to want to know about this. We are going to be talking about artificial intelligence and the applications for AI now, more specifically in eyecare.
Keri and I are going to talk about what is going to be applicable to people in any industry. I have my wonderful guest, Keri Sculland, who is the Manager of Editing and Content Strategy at Marketing4ECPs. Keri has spent over twelve years in writing and editing, first in newspaper and magazine, and now doing her digital marketing. Prior to this, she also worked in a big busy optometry clinic in Calgary, a multiple-location clinic where she had a lot of hands-on experience in marketing and patient experience and that type of thing. She’s bringing a lot of cool insights from these different perspectives all into this conversation. I’m excited we are going to be talking about how we can start to use AI now and how it’s applicable to us in our work setting. Thank you so much, Keri, for being here. I’m excited to have this conversation.
Thank you, Harbir. Me as well.
Before we had this, we’d been chatting a little bit. How should we have this conversation? How should we approach it? The idea was like, “How can we apply AI today?” The answer to our questions of how to approach the conversation was answered by none other than AI. You did a great, funny little thing, which is ironic and convenient. You asked your AI platform what questions should a podcast host ask an AI-assisted content creator about AI and optometry, which is the perfect prompt to get the perfect question.
The questions we are going to be asking in this conversation are from AI itself. You use an application called Jasper. We’ll get a little bit more detail about ChatGPT and what Jasper is and some of the other applications out there. I thought this was the perfect way to have our conversation structured. The first question that Jasper offered you was, “How can AI technology and content creation tools be used to improve patient education and optometry?” What do you think?
That’s such a good question to get us started. AI is great for assisting in content creation. It crawls the internet. It does a lot of that research for you. The outputs that it can provide you are simple to understand. They’re typically factual. I do always recommend fact-checking everything that you get because AI is only as good as the content that’s already published on the internet. It takes that first step out. There are a lot of ways that you can use it similarly to the way that you would on Google, “What is an OCT?” It’ll tell you exactly what it is and you can ask who it’s for. It’s great for education and content creation.
The content creation thing is something that we probably have multiple verticals to go through multiple chapters of this conversation in that one thing. I was saying to you before, “How do we define AI?” How would you define AI? There are a lot of smart applications already. Specifically, when I get in my car and my phone connects to Bluetooth, it predicts where I’m going based on my pattern throughout the weekend. My schedule’s pretty predictable. My watch will beep and it will say, “It’s four minutes to your kid’s preschool,” because it knows at 8:30 in the morning that’s likely where I’m going to drop her off at preschool. When I get back in the car after dropping her off, it says, “It’s 30 minutes to your work. Traffic is moderate. You can check the map if you want.” It knows where I’m going and what I’m doing. Is that AI? What’s the difference? Where do we draw the line, and how do we start calling it AI?
Honestly, I feel maybe we’re not using the term correctly, but it is one of those buzzwords. AI is something that everyone understands. When we look at something ChatGPT, the GPT stands for Generative Pre-trained Transformer. It uses language models to give you the outputs that you’re expecting. It’s the same thing with your commute to work. It’s gathering data about you, and it’s giving you outputs based on that data.
I don’t know if it’s technically artificial intelligence or if it’s a good formula that understands what you need at that moment. A lot of the content that we’re using is only as good as the inputs that you give it because it’s using content that already exists. The same to you with your car telling you how far it is to go. We have things like Alexa and Google Home, and people have them in their kitchens, “What’s a good recipe for this?” and it will tell you based on preexisting content that’s already on the internet.
I felt this was almost a fancy extension of Google. The information is there. It’s not technically making stuff up out of thin air. It’s taking information that’s there and presenting it to us in a more digestible way or in a format that is specific to what we’re requesting. GPT, this generative stuff, is not AI in the way that we imagine AI in iRobot, where there’s one big brain that controls a bunch of robots and can overtake us.
No, I don’t think it’s there yet. Maybe one day. It’s pretty cool that we can ask you questions and it can search for answers for logical things that we’re looking for.
It certainly can save us a lot of time. I’ve been goofing around with some different applications. I want to talk about the content creation thing. What are you using it most of the time? Are you using it daily? A lot of your stuff is through this or to some? How much are you using it?
We’re using it often. It’s important to differentiate that we are calling it AI-assisted content creation. We’ll have all of our writers going through it, tweaking the content, and making sure that it’s hitting all of those regulations, it sounds human, and we’re making changes to ensure that we’re getting a brand voice for all of our clients, and that stuff. It’s pretty interesting in the ways that we’re using it. We’re always finding new little ways to use it. It’s cool because it’s allowing us more time to work on the branding and the messaging.
We’re not sitting there thinking, “How would I craft this message?” We can type it in, it gives us an output, and then from there, we can work on, “How do we make this exactly perfect for this client?” We are using it to help us generate some social media captions because we generate a lot of those in a very short amount of time. We use it to get us started on blogs. It gives us a good framework to work from every time. We are never publishing AI generally.
That’s important. I know you don’t want your clients to be like, “These guys are sitting around punching stuff into AI and then eating a sandwich.”
There’s a lot of work that goes into it. I would highly recommend having a second set of eyes on it at all times.
I understand other eyecare professionals are thinking along the same lines as me. We’re thinking big splash type of stuff like, “How is this going to change the entire way I practice?” Now, there isn’t anything that’s going to change the way you practice. I’ve been talking to a few ODs and a few other people who are in the tech space. There is technology, sure, maybe on the cusp. There’s probably the technology that we are utilizing, but it’s not totally changing the way we’re practicing. The amount of content we need to start putting out there, and we’ll continue to need to put out there, whether it’s emails, blogs, social media, and the fact that we have these little engines that can create a lot of that content for us can save us a lot of time.
It can help us get people who are doing a little bit of content creation and are limited or not doing as much. They don’t feel they have the time, or perhaps they don’t feel they’re creative enough. It definitely can get you over that hump pretty quickly. If I can give you quick example of what one thing I was testing out was writing and creating email templates for patients. I went to ChatGPT and typed in, “Create an email template from an optometrist to a patient explaining the benefits of omega-3.”
It worded it fairly well because it was towards a patient. I then said, “Clarify this sentence. Change or expand on this sentence. Specify EPA and DHA content.” It will give based on information on the internet. It crafted a pretty good email that I’m likely going to use after making some small edits. That was the very first thing I did on ChatGPT. It took me a little bit longer because I was trying to figure it out. That could have been done within five minutes or less. There are a lot of applications in that way too.
It sounds like you did maybe a little bit of research on the prompts that you need to give it. Those were all keywords that we use.
I guess you could say I did research. I’ve been watching other people and listening to what other people are doing. There are lots of stuff out on social media about how to use it. I wasn’t digging too deep. That didn’t take me a lot of time. Anybody reading this can take the words that I used and go and that’s it. You need to be like, “It’s from an optometrist to a patient on this specific topic. Make it this length.” You could use whatever prompts or when it gives you the first one, then say, “Make it shorter.” That’s what I did.
It was too long at first so it made it shorter. Now, “Give me more information on this topic.” You don’t have to go back and be like, “Change the fourth word of the sentence.” You can say, “Tell me more about this thing,” and it spits out new answers. It’s pretty cool how user-friendly it is. I’m new to this, so it’s exciting for me and it tabs it. If you start one conversation on the email about Omega-3s, it tabs that conversation. If you start a new one like, “Make me a presentation on something,” it will tab that. You can always go back and see your history on all of that. It’s very easy to use. In your opinion, social media-wise, what is the application? How would I go about even creating a social media post using AI?
Social media is a cool and unique one. I like AI-assisted content for idea generation. Whatever AI you’re using, “Come up with ten ideas about myopia control that I can post to social media,” and it will come up with a few ideas and then you can work with that. It’s a great starting point. That’s how I think AI content generation is a great point to kick off from. You then can dive more into things and use your own personal knowledge to build out these ideas that the AI has created for you.
How about on the caption side of things? I know that’s where a lot of people get stuck to like, “I got a picture now. What do I say about this picture?” How can we use AI in that sense?
You can put in a description of what the picture is about. You can put in other prompts to come up with a social media caption. It’s pretty good. I always do say that I think that you should have someone who has content experience on this because we’re talented at making it human and making sure that it matches your brand voice. We run the risk here of coming across as inauthentic if we’re, all of a sudden, posting things that sound strange and not genuinely you or us. There are some things that maybe don’t live up to the regulations in your providence or state. That’s something that you need to be cautious about.
It’s very important. I got it. How about on the blog side of things? You were mentioning you guys do use it as far as creating the meat of it and then you edit it from there. Is that how it goes?
We’ve been using it to get that outline out there. It takes a little bit of the research out of it of, “What am I going to put in this blog?” It uses smart words and then you can add keywords and that thing. It’s such a good starting point. It’s very similar to social media. It will give you something to work with and then you can judge it from there.
To echo what you were saying, I agree with that. A lot of times when I’m making a presentation or writing a blog or something like that, I’m like, “What am I going to even say? How do I even say this?” If I have something in front of me, even if it’s not very good, I can look at it and say, “I can understand how I can start. Maybe I’ll change this, scrap that, or reposition this.” It’s a lot easier than a blank page staring at you and you don’t know what to write at all. I’m working on a presentation now to talk to ODs and ECPs in general about these applications for AI.
I was talking about this before. I wanted to go to AI and say, “Give me five slides on how AI could be used in eyecare.” I wanted to use that as part of my presentation to show people like, “Here’s what AI did for me.” In ChatGPT, I did that. I said, “Give me five slides for a presentation on how AI could be used.” It gave me a pretty good overview like, “You should talk about this and then you should talk about this.” I went and said, “I’ve me specific information on this and this.”
It was pretty good. It broke it down into an intro, the history of AI, the current uses of AI, future uses of AI, and a conclusion of why we should all be or how we should all get into it. It was good information, but the actual creating the presentation part, like putting it into a PowerPoint slide that’s visually captivating and all of that, that part’s lacking now. At least it gave me the content when I’m making a presentation. That’s the hard part for me. It’s like, “What do I even say?” It gave me the content to put into those few slides. That’s cool. It is a starting point.
Let’s go back to the AI-assisted questions that were crafted for us. We’ve been going off on different tangents here. Let’s talk about ethics a little bit because that’s going to become a big thing, intellectual property and things like that. If ChatGPT puts out a blog for me, how do I know I’m not copying somebody? We see AI-assisted music being made and using Drake’s voice or Eminem’s voice to create a new rap song. It’s like, “The lyrics are new, but the voice is somebody else’s.” I don’t know what the IP laws are around that if there are any. What do you see as far as ethical concerns? How should we approach this so we’re being aware of that at least?
A great place to start is with a plagiarism checker. Anything that you’re getting out of these generative predictive outputs, you should be running through a plagiarism checker to be on the safe side. Having a look at the content that it’s putting out and changing it enough is important. You need to have eyes on it. You can’t guarantee that it’s even putting positive things out into the world. Sometimes it seems it might make a fact or two up. You need to be careful about that.
It’s getting human then, basically.
You can’t guarantee that everything that exists on the internet currently is factual and correct. It’s pulling information from places across the internet. That’s where we run into the issue too of, “Is this content up to advertising regulations? Is it going to get you in trouble?” You need to ha be having a good look at it before you go and post it anywhere on your website, social media, or anything like that. The plagiarism checker is a great place to start. When we start talking about the voice AI, that stuff, first off, it is creepy. This is not my thing. I was at a conference and they created a video that used the gentleman’s face and his voice.
You can’t guarantee that everything that exists on the internet currently is factual and correct.
They recorded a minute of his voice and then sent it through an AI voice creator and put it on with this AI video of himself. It was weird. It was a little uncanny valley-esque, that’s to say the least. That runs us into ethical concerns there. If you don’t have consent from someone to use their voice, that could get you into some hot water. It also runs the risk of coming across as inauthentic and creepy. I would play cautiously with stuff like that.
There’s certainly a creepy element to it because it’s almost real, but you can tell it’s not real. That makes you feel a little uncomfortable about it. On that note, it’s part of my goofing around with these different tools online. I created a little AI. I had a photo of myself when I was 4 or 5 years old. I put that into the AI thing and I made it say fifteen seconds of me saying the importance of a comprehensive eye exam.
It’s this little child that you know doesn’t know those words, but it’s coming out in a child’s voice. It’s super cute, but at the same time creepy. The comments under the post, if you go and check it on Instagram, most of them say that it’s creepy but cute. That’s a static photo, but the head’s moving, lips are moving, and the eyes are blinking. It’s a little odd, but it’s there.
I did watch that. I thought it was pretty funny. I feel it’s not that far off from Snapchat filters and stuff like that.
The funny thing was the day before, I posted that my cousin sent me a video of my niece, who’s eight months old. It was some rap song and her mouth was moving to the lyrics. It’s super cute. This little baby with a chubby face and a mouth is going off rapping. It was basically that. Snapchat already has that. You can create filters of yourself, but it’s going to be specific to the songs or whatever that are on the app. Whereas these other apps will let you make that face and say whatever you want it to say, which can be cool or it could be bad.
Like any tool, it could be used for negative things. You get into the potential of people using other people’s likeness and voice to get them in trouble, even possibly. You want to be careful about not using somebody. You don’t want to take Prince Harry’s face and voice and have them talk about why everybody should come to Clarity Eyecare or something. You might potentially get in trouble.
You do run the risk there. I’m sure there will be some legal suits in the future. I want to circle back to the fun aspect a little bit because it did spark my memory about trying on glasses online. There are some online optical retailers where you can take a picture of your face, move it to the left, move it to the right, and then the next thing you know, you’re trying on glasses. You’re moving your head around and seeing how you look with them on. That’s a cool application. I thought of that because there was a Snapchat filter where I could try on glasses. That was extremely clever.
Weirdly enough, Snapchat is better at it than most of the other applications that an eyecare provider would be able to purchase or invest in to bring into their office. Snapchat filters are good, but we’re talking about a multi-billion dollar corporation investing in their own tech versus somebody else. That augmented reality type of stuff is going to become important for us and eyecare. For those of us out there who are interested in like, “How’s AI going to change the way I practice?” I’m going to be having more of those conversations coming up, lining up some guests to talk about what tech is in the works that might change the way we practice.
Maybe VR-type stuff is going to be applicable in optometry. Whether it’s talking about refractions or some other ocular evaluations or trying on glasses, there are going to be applications for all of that, like retinal evaluations, dry eye, and all of this stuff. We’re going to talk more about that. It’s just that there’s not something tangible, impactful right now. I’m wondering, Keri, if you have any thoughts in the case of some of the more backend type of stuff, like scheduling or communicating with patients. Do you have any thoughts on applicable uses for AI there?
I get excited about this because this is a cool part of the patient care customer experience. There are AI content creators that have browser extensions. That makes it easy if you’re viewing email that way. It can read through the email that you received and if you want to come up with a response that’s like, “How do I tell this patient that we have their contact lenses in, but we still need payment from them.” It’s something like that. You can be like, “Write an email about this,” and it will take off and write you a well-worded personal email back to this patient. It is letting them know that their order is in. There are also different applications too that you can use it.
I don’t know how many patient recall emails I’ve seen that are not great, but they could be better. It’s something that you generally, as an optometrist, set it and forget it. You know that it’s sending them the email at the appropriate times when you’re scheduling software like, “Here are all the people who are due for eye exams in September.” Maybe it’s worth having a look at those emails and sending them through some generative predictive text and getting them refreshed, nice and friendly. Maybe convert those recalls into leads again and get those people back into the office.
That’s great. Right off the top, when I introduced you, I mentioned you’ve worked in an optometry clinic. It’s a busy clinic, and you were doing a lot of hands-on with patients. It’s cool that you could bring that insight into the conversation. That’s that tangible day-to-day stuff that we start to get excited about like, “All the cool tech that I could maybe use as AI.” This is stuff we’re doing every single day and can impact the business.
We tend to overlook it because it’s mundane. If we could revamp that and give the patient a different experience in that communication, that’s important. For those of us who are very new or have limited or no experience in the GPT type of stuff, what I would do step by step if I wanted to revamp those recall emails? Let’s say I’ve got ChatGPT on my computer. Let’s say I figured out how to do that much and I’ve logged in. What should I prompt it to say?
There are a couple of different ways that you could do it. You could copy the emails that you have already and get it to rewrite them in a friendlier tone or something. You could start from scratch. It’s up to you.
I have emails I’ve been sending for years. They need to be revamped. I’ll take it, copy the text, paste it into ChatGPT, and maybe above that, “Rewrite this following email in a friendlier tone mentioning our clinic name,” and it will give me some. The good thing is there’s no wrong way to do this. You can’t make a mistake. If you prompt it incorrectly and you don’t the response, you can say, “Let’s start again. Let’s try this way.”
If it gives you a response, you’re like, “That’s close but not quite there,” use words that you’d use when you’re talking to a human being. “Change the last sentence. Make it sound friendlier,” or something like that, and it will start to make little changes and it becomes fun once you get that feedback from it. That’s a cool application to help improve that patient experience. Do you have any other thoughts on that as far as improving the patient interaction experience with the clinic?
The chat functions are cool because if you’re sending text messages to patients, some people send personalized text messages. Those probably take a bit of time to craft. Maybe you could come up with a handful of them that are reusable content, like someone has an exam coming up in two days and you want to confirm, or you want to let them know that their order’s in or something like that. You can reuse that content but send it through the chat and see what it comes up with. It’s cool because it takes that initial like, “What do I say to this patient?” out of the equation. You can get the chat to send you something. You check it over, make sure it’s correct, make sure it’s friendly, and it’s what you want it to be. You can send it off as long as it’s factual.
It helps you get it off the ground. “Craft me a text message to remind a patient that they’re scheduled for an appointment tomorrow.” Be specific about what type of message you want it to be so it will know, that their patient’s contact lenses are ready to pick up or whatever. All of a sudden, you got five text messages that could be reused and maybe personalized with the person’s name or something. That’s cool. What do you think as far as the future of all of this? Where do you think we’re going? How do you think AI is going to start to impact eyecare with the different things that you see on the horizon?
There’s a lot going on in the “AI world” now. The future is bright. The future is AI. We should be getting into it now if we’re not already. There are so many ways it is going to be able to help us and streamline some of the day-to-day things that take us a lot of time. One day, I could see using AI or computer program software to read through a patient’s images or OCTs and make recommendations based on the little blips and findings that you get in those images. It could be reading through patient files and making recommendations based on their health and their family history. It could help an optometrist come to those diagnoses, double-check them, and provide in-depth care for their patients.
The future is bright. The future is AI. We should be getting into it right now if we’re not already.
I don’t know if data mining is not the right word or phrase, but going through that data that’s for us, “We’d have to flip this page, flip that page. I got to go back to last year’s chart. Let me see.” It is being able to go through the data and come back and say, “Here are the changes, trajectory,” or whatever. Creating these predictive models sounds exciting too. It certainly can help us be better clinicians. That’s cool to see on top of the other ways it’s already possibly going to help.
There are going to be some cool applications in the medical field because doctors use a lot of the same acronyms, especially in optometry. You’re using BID to tell people how often to take their medication. Any software that should be able to read those acronyms and understand exactly what you wrote and what they were doing a year ago. You’re totally right in terms of looking back on patient files and their treatment plans.
That’s very cool. I want to tell a quick story about a friend of mine who’s a lawyer and echoing all the stuff you’ve already said. He’s a bit more advanced because, with OpenAI, you mentioned it’s open source code or software. Pretty much anyone can take that and change it up into using it specifically in the area that they need. For him, if he gets an email from somebody at another lawyer perhaps asking for information on a client or on a case, he could prompt it. It will go into his database based on this secured server he’s got and pull out the information based on that question that the person had asked versus him sending his assistant or him going in there personally.
It will basically bring up the information, then he’ll say, “Write me an email response. Add this as a file or add this as a blurb in the email.” It’s pretty cool. For somebody with lots of words to read through constantly, I imagine it could save him a ton of time. We are not reading through as many words and documents but still going through other forms of documents, imaging, or dry eye information. It’s exciting to see how it might be able to pull that information and get us to those diagnoses or treatment plans a lot faster.
I’d be curious to hear what software your friend is using. That sounds exactly like it could be used in the medical sphere.
Once he told me that, I’m starting to email people around eyecare and the EMR and EHR type of space, “What have you guys got cooking now?” I’m interested to see how this could be applicable. This might help me make a decision on the next electronic health record system that I implement in my office. If it’s going to help me access information more quickly, that’s always easy. Otherwise, trying to go through ten years of files to find an answer to a question is a pain. It could be helpful. I’ll ask him and I’ll get back to you. Any other final words of wisdom, information, or exciting things you want to share about AI before we wrap up?
AI is super cool. Everyone should be at least playing with it and seeing what it can do and how it can help them in their day-to-day life. We use it in some cool ways already when we ask Alexa or Google our questions. The only thing that I want to stress is that it needs a close eye on it. Never blindly post anything that AI gives you. Have a look at it and make sure it meets those advertising regulations. There’s a lot of them for optometry. Make sure that it’s friendly. In the end, you may need to tweak it to make it sound authentically you. It’s such a good foundation to build off.
AI is super cool. Everyone should be at least playing with it and seeing what it can do and how it can help them in their day-to-day life.
That’s great. Get excited about it and get involved in it is the key thing as well. If we sit back and we say, “I don’t know about this AI thing,” then we’re going to have to start playing catch up. Now is the time we start getting involved. We’re not necessarily ahead of the curve, but we are rolling with the technology as it is. When something changes, we’re able to adapt to that more quickly rather than trying to climb up some mountain of all sorts of new developments that we didn’t keep up with. Get in there. As my friend Shawn Kanungo says, “Get your hands dirty. Get dirty. Get in the mud and play with it.” You’ll be pleasantly surprised at how helpful it can be.
That’s exactly the right attitude.
Thank you so much, Keri. I appreciate your insight. I’m excited. I’m going to take a lot of what you said and start to fool around with that a little bit more over the next little while. I hope everybody else who’s reading does that too.
Thank you so much.
Thanks, everybody, for reading. Make sure you share this. Again, take a screenshot or send a link to somebody. If you think that there are people out there like friends of yours that you should be getting a little more involved in AI, let them know that Keri and I had this conversation so we can all get up to speed and have fun with this technology. I’ll be back with another episode very soon. See you then.
