Welcome to Real Talk on AI and I care with your host, Dr. Scott Morris. And I'm co host, Dr. Rehan Ahmed.
In each episode, we discuss current news, adding a little AI education, and debate innovative AI tools and topics that will change the industry. Well, hello, everybody. Welcome to Real Talk. Today, your host, I'm your host, Scott, Dr.
Scott Morris. And today I have on Fiona Buckmaster, who's out of the UK, and she's been doing some amazing work on AI in education. How do our students spend our professors and academia around the world, address AI, and what do they really think as they're talking about it? So before we get started with Fiona, I always do the knowledge bite.
So we're going to talk a little bit today. Here we go. Today in our knowledge bite, we're going to talk about micro learning. And as I just did my introduction about Fiona, we're going to spend all of our real talk time kind of learning about the future of AI and education.
I wanted to talk a little bit about how micro learning is kind of shrinking the classroom. So when you think about the last time you learned something new, did you go to a three hour seminar? Or was it a little two minute video on YouTube or tick talk about how to fix a leaky faucet? If it was the latter, you've experienced what's known as micro learning.
This is the practice of breaking down complex information into bite sized little nuggets or pieces, usually between one and three minutes, designed to meet some specific learning goal without just overloading you with all kinds of information you don't need. Now micro learning is not new. Artificial intelligence, though, is turning into kind of a let's call it a superpower. Traditionally learning these little mini lessons took weeks of design.
And today AI can ingest just massive amounts of information, textbooks, videos, and instantly distill it down into a series of short little summaries or interactive quizzes. But the real magic is AI is now personalizing that education. It's acting as a digital tutor that knows exactly what you've forgotten or what you never knew in the first place. Instead of everyone learning the same five minute clip, AI soon will analyze your performance and serve it up into let's call it a specific bite of info you need right now.
Kind of like what we do with knowledge bites. It's moving us from just in case learning to just in time learning new concept. So why does this matter because of what's called the spacing effect our brains contrary to contrary believe contrary to common belief are not sponges their filters. We forget 90% of what we learn within a month if we don't revisit it in some lectures.
I think it might be by the time they walk out of their door. AI driven micro learning uses algorithms to trigger what's called spaced repetition. It pings you with a tiny little lesson right at the moment your memories about the fade. And by hitting these little small intervals, we move information out of short term memory and more into long term memory or better yet maybe mastery in a world where attention spans are competing with a million pings.
Micro learning powered by AI isn't just convenience. It's how we're going to keep up. This was your knowledge bite. Let's get on to Fiona.
Now I'm super excited to get to Fiona. Fiona at first of all, thank you so much for being on the show. I really appreciate it. Fiona is.
She's got her PhD at the University of Huddersfield. And she's working on some really amazing stuff and I will let her do the justice of telling you what she's working on because I'm sure that I won't get it all right. She's got to see also works for the Brian Holden Foundation as kind of an automatic education and development consultant. I'm sure she's going to weave some of that job responsibility into what we're going to talk about too.
So Fiona, welcome to the show. Super happy to have you. Thank you for agreeing to join us. Thank you very much for inviting me.
Hopefully everyone can understand this Scottish accent. Oh, let me just tell you one of the secrets. We love your accent. It's never going to be.
So Fiona, tell us a little bit about yourself and what you've been working on the last couple of years to kind of set the stage for what the audience needs to know about you. Sure. So I'm an optometrist and educator. So I work as a primary care optometrist in Scotland.
And then education, I do an amix of different things. So a lot through the Brian Holden Foundation. So I work with emerging optometry schools to help them develop their education program. So that's primarily been inviting them the last few years.
I got into the AI space in around 2020. So I worked as a researcher on project called school, which is the Scottish collaborative optometry optometriop and technology network for e research, which is why we use the acronym because it's a bit of a time to answer. And I was basically the link in that project between the data scientists and the people sort of developing AI for written image analysis and the clinical practice site. During that, I didn't really know much about AI before then.
And so I had to learn a lot and experienced that there weren't that many resources that were particularly suitable for me as an optometrist coming into this space. And then we're recently transitioned through my PhD work. So I don't have my PhD yet. I'm still a PhD candidate, but my PhD work at the University of Huddersfield where I try to figure out how we integrate AI literacy into undergraduate optometry programs.
Well, so what's the answer? We'll cut right to it. So, you know, because I, and for the audience, you know, I've been on, Fiona and I've been on the same mission for the last her a little longer than me, but like I have always believed the way to the future is to start with the younger generation who's learning and integrate what different technologies in this place, AI, aren't to not just their everyday clinical experience, but into the way they learn. And so, Fiona, you know, whenever we face, whenever we try to implement something new, we always have to get over the single biggest challenge, which is the fear of change.
Right. And so what do you see in your research? What do you see as some of the biggest challenges that we have of integrating AI into optometric and ophthalmology education. And where are we at in terms of that process?
Like in terms of like the innovation curve or are we really still at the very beginning or are we, you know, what's the battles we're fighting? Yeah. So I guess there's sort of two prongs to that. So are we talking about AI as an educational tool?
Are we talking about teaching clinicians about AI and their clinical practice? So that's both. Yes, both. Two different ways in both.
Take one at a time. Right. So how about how about we start in terms of how do we use AI and or other innovative technologies to better educate our future generation of doctors? Yeah.
So in most sort of contemporary healthcare, health professions, university programs now we're now using something called a spiral curriculum. So you teach something and then you kind of come back to it the next year and add some more knowledge. You come back to it again and add some more knowledge as you go on, you're kind of building that spiral. It's going up and up and up as they gain more knowledge.
So the education theory that unoppins this is called constructivism. So we're constructing sort of a world view that our learners hold about their knowledge by repeating and coming back to things and adding more each time. And AI tools, if we're thinking about generative AI, being used as a virtual tutor, that's inherently constructivistic can learn what our learner knows and then give it to give them tailored advice on where they need to be adding more stuff and. Before you go on, I think that's interesting because you know that's one of the challenges I've always faced from the podium.
I've never taught an academic institution, but you know I've taught thousands of hours from the podium is that the question is always just what you bring up so we teach and we test. And then we leave and not only do we have to assume that providers are going to remember what they learned, but what about the stuff they didn't know in the first place, the gaps that were there. And I'm very curious. I want to probe into that a little bit about how does AI look at gaps in knowledge and fix those if we don't need to relearn what we already know.
So we need to learn the stuff we don't know and our current system is really not built to do that. And so I'm curious your thoughts on that. Yeah, so it's very difficult because you don't know what you don't know right. So if you're a practitioner, you're doing some continuing professional development, it's hard for you to seek out things that you don't know about.
In the same with students, they will have gaps and their knowledge that they're not aware of everyone does. There are some specific systems being developed for this. There's some work at Cardiff University and Wales been done on a kind of intelligent tutor system that's specific for optometry. But we already know that students are using freely available generative AI tools like chat GPT to make them some flashcards or things like that, even if it's not the kind of other academic implications that every educator is worried about.
We know that students are using these tools already. And so I think as educators, it's it's useful to try and understand how are they being used. What is my institution's policy on the use of these tools and kind of leaning to it. We're kind of past the point where you can ban it.
It's not working. I just think that it is I look at my kids. My kids are, you know, I've won the just graduate engineering school and get rid of law school and I won that's in his junior year of college. And it's just a tool for them, right.
It's no different than I'm much older than the fion of it. You used to go to the library and pull out a book and write a note card and go, these are my cheat sheets, right. Now they're just skipping the library and skipping the go to and skipping the note cards and going, well, here's how I learn. But that's still the old method of learning.
Yeah, so I think what's probably useful is for educators to learn how we can integrate AI without having it be a bypass to that learning process. Don't just tell me the answer because actually the process of us searching for the answer often brings up a whole bunch of other stuff, which is useful to know. I spoke to other things. Yeah.
Yeah, I spoke to a clinician recently who said that he felt that Google had become an answer engine and not a search engine anymore because it gets you little AI overview at the top. But it's so true. What do you find? What did you find on page two or three of Google results or whatever when you're trailing through trying to find some information.
You often find things that you didn't know you were looking for, whereas a generative AI tool maker, if you're a really nice summer of the question you asked, but it's not going to summarize the question. You didn't ask. And it might even be right. Yeah.
Yeah, so I think it is a space that you obviously, if you're an educator, need to be working in line with your institution's policies on AI usage. But it's useful to know how they're being used and how we can use this to maybe optimize or help your searches so that you're not just asking AI for the answer, but saying, how would I search for this thing that I don't know about where should I go to look for primary sources of information to help add into, you know, just spitting out an AI summary. Well, so, so fill me in. You know, it's part of your research.
So that's all great theory. So what's actually happening like what what are professors at least in the UK. And I'm assuming we have the same challenges with the professors here. And what is useful?
How are they actually using it? Like, you know, we can sit there and give it the 30,000 view. This is what it should be. But I knew you're doing research into what's actually happening.
What are they actually doing? I'm curious because I want to learn what are my students that are coming through. How are they actually learning what I'm asking them to learn? Because I don't think it's the way I learned.
Yeah, and I think that's that's probably fair. And in the last sort of two, three years, the way that we're teaching has changed so much because these generative AI tools have become so widespread. And so a lot of educators and institutions are on the back foot about they didn't anticipate this change coming in. A lot of big institutions don't change things very quickly.
And so this, the students are. Yeah, they change your generations. Not days. Yeah.
So the students are outpacing the educational institutions in a lot of ways into how they're integrating this into their practices that, you know, that's a generalization. But there is something of a view that universities or higher education providers have a negative view of AI because of things like the risk of plagiarism, the risk of what we call cognitive offloading. So you offload a task onto tool and then you become worse at doing that task because you're not practicing it. Like I used to know all my friends, phone numbers when I was a kid.
I think I know one phone number. And I off the top of my head, not even sure I know mine. Yeah, exactly. So there are all these risks that come along with the use of AI tools and education.
And so that can lead to. And I think it. And, you know, I think about that, but I was around when the internet started right or quote unquote became popular. You know, and that we are having the same discussion back then, you know, I'd been 10 years, 11 years out of school at that point or something like that.
And, you know, my friends who were teaching the academic is all this internet thing. You know, it's going to change the way it's going to room learning. Well, it didn't room learning. It now, like you said, you go to what did you call it, not Google search engine, Google answer engine, right?
Is that you, you queried it to get an answer you were looking for you search and answer. Well, I, you know, this is just the next evolution of the internet. But now instead of search, you're prompting, right, different different terminology, same basic idea. And so, what I think is that I agree with you, you know, there's all those risks of hey, let's offload stuff and you know, can't really some stuff probably should be offloaded because it's just busy work.
But it's interesting to listen to like my sons or listen to some of my student doctors of how they use it. They ask questions that I never thought about asking when I was their age. So is this tool going to there's all those risks. I agree, plagiarism offloading, not actually learning, but just getting the answers.
But, but is AI going to teach students humanity, not just students, but humanity to learn in a different way. And, and number two part of that question is, is that how does academia embrace that or are we going to undergo a fundamental change in the way the educational system works, not to put you on the spot, human. Yeah, do I think the entirety of education is going to change probably yes, we do need to change the way we're doing things so that we're getting the benefits of these tools. It is a personal I it can be a personalized learning assistant.
You know, if you have a student, so 200 students and your lecture theater, however many it may be, you're not keep you're not able to keep tabs on. What they're understanding or not understanding. Now you can have an assistant do that for you. So that's a great innovation.
But, you know, we will probably have to change the way that we integrate things like assessments and things that like that. And so that we're not just trying to ban AI or create assessments like moving away from written coursework because it's not going to be a great way to change the way that we integrate things like assessments and things that like that. And so that we're not just trying to ban AI or create assessments like moving away from written coursework because it can just, you know, put a prompt into a generative AI tool and get something that reads sort of read. But actually help students to maybe interrogate the outputs that they're getting from AI tools learn how to critically think about the things that they're being presented with so that they learn how to use these tools in a way that's going to benefit them.
Yeah, I think we were doing a podcast the other day we got talking about, you know, I think we're undergoing a fundamental shift of, of healthcare providers from being data gathers and in data interpreters to now being. Tech now and we're going to go to move into the second part of the discussion of how do we teach students to use AI in their everyday clinical practice as well, but I think we're moving from where we were data gathers and then data interpreters to now where we're data interpreters data translators and change management experts right so it's a different set of knowledge or skill sets that we maybe need to teach because. None of us have the cognitive of in my opinion I'd be curious to see what you think none of us have the cognitive ability of an LLM in terms of remembering everything, but how do you adjudicate that knowledge and cause transformational change in patients to provide better care. AI is not going to be able to do that anytime in the near future, but so what are your thoughts on that we were where do you lie on that.
How are our roles changing. Yeah, so you know I firmly believe that AI is not going to replace the optometrist, but I also believe that our job is probably going to look very different in 20 years than it does now in the same way that our job looked different 20 years ago than it does. Very different. So really it's the place of the optometrist to be that human connection and particularly when it comes to someone's health, I think patients are looking for a human being to give them that reassurance in that empathetic connection.
And one of the things that I find quite exciting for me personally as a clinician is not even what it can do to my diagnostic capabilities or something like that, but can it take away some of my admin burden I do way too much paperwork can I focus on my patients and not on all this other stuff that I don't really want to do and can't bother doing but I have to do because it's now part of the job. I agree, I mean we'll operational AI give us the time and can't do the hopefully more energy because we're not frustrated doing the stuff we don't want to do as you said, well operational AI give us the time to better use diagnostic AI to provide better care and I agree with you and I think the answer is that yes, I mean I think we're still in the early stages of that but well, well, being we're doing what we do actually be fun again. I hope so right and you can't see that I'm actually doing this on a video call Fiona and she's got a beautiful smile and literally her ear she just smile ear to ear on that one so I know I struck a chord on that one she's got this big beautiful smile right now so well tell me your thoughts on that project like how how where do you see AI helping change how we deliver care so we're moving kind of from the education to how we teach to deliver care and we're going to come back to education in here in a minute of how we use that to educate patients too but how do you how do you see in your experience and what you've done your research and especially going to start these these schools what are the tools that are going to be most impactful for truly delivering better care. Yes, so I think when we talk about AI and I care there's a lot of talk about efficiencies and improving efficiency and I think this is certainly possible but a lot of people talk about the efficiencies of an individual practitioner or an individual practice so you're going to be able to do a faster I exam or you're going to be able to see more patients or something like that and yeah that's probably all true.
But because I'm a big public health nerd I like to think of a systems wide view I work in in Scotland we have a national health service which is what our I care is provided under and so we're not just looking at an individual practice being efficient but the health system is a whole being efficient and this is the same and developing countries and things like that you have a very limited health budget that you need to use appropriately. And so actually a lot of the efficiencies come from not an individual practice or practitioner being quicker at their job but making sure that patients are seeking the right care at the right place the right part of the health system at the right time so someone who needs to be in the ER is in the yard but someone who doesn't need to be in the yard is not in the ER because an AI chatbot has helped to triage them and send them to the most appropriate place or even from the treatment side of things. If we have personalized medicine and we can know what treatments are going to work best or predict what treatments are going to work best for a patient we don't waste time and money given the treatment that's not going to work for them. If you scale that up to an entire health system that's a lot of savings in terms of time and money and all other kinds of efficiency so it's tempting to focus on the individual practitioner view and you know I've got skin in the game in that one I'm a practitioner I obviously want my day to be as efficient as possible.
As frictionless as possible. Yes, exactly. Well, I think you brought up a great point there that I think so an audience this is one of those about once once a week when we do this I say pay attention right here and I think if you want to just hit on an awesome incredible point that I just don't think we hit on enough is what happens pre exam right if how do we know because there's a whole bunch of the care delivery service. So a bunch of the care delivery cycle that happens before a patient ever makes a phone call or goes online to a portal seven point when they're going it is a problem I need to worry about is this a problem I need to see somebody for can I take care of myself if it is somebody who do I go see how do I know they're good how do I know they're going to solve my problem and how do I find them and are they somebody I can get access to and there's a whole bunch of decision making patient or healthcare consumer decision making and if you any you lay that out beautifully maybe the best of anybody I've heard is that that's going to change the economics of an entire healthcare delivery system so instead of a patient to open an ER which here in the states is in the states is you know five times more expensive to show up in a primary care clinic clinic in sedation of an ER could they go hey I can wait a day you know to get in and see somebody and that changes the economics of of a country of of a world you know global economy I think that's a fascinating observation that audience I think you need to pay attention to is this is where AI and a chatbot and or an LM or whatever and I definitely think we're early on in the spectrum of that.
But it's going to change the way we do it and then like you said Fiona on the backside of right now we are reliant I call it the end of one Fiona is all of the clinical decisions I make are based on an end of one my brain and I'd love to say my brain keeps up with all 600,000 pages written articles that come out every year and I care I don't you know will AI be able to give us the most current up to date access to tools and evidence based medicine. That focuses on phenotype genotype of individuals and you as you said you guys have a national system so you get to look at records in a very different way that we do in all our silo D.H.R.s. I just want to compliment you I mean that whole beginning before the patient ever shows up on the door something we so often just ignore it's a really important point so that brings up how do you think that AI is can impact kind of an extension of that how do you think AI is can impact clinical care for you. Clinical care from a consumer perspective how they receive care much less fine care.
Yes so as I say we expect an AI to impact every single part of the I care patient journey so as you see from before they ever decide to reach your practice until they're at home and maybe getting some at home monitoring or something like that things like that becoming more possible. I think for me obviously I'm a community I care practitioner so I really want to keep patients in the community as much as possible I think being able to get good care in an environment that the patient is able to access and is comfortable with is is going to be a benefit so if they can stay in the community or stay in their own home whilst receiving I care. You know it's someone that's local to them it's not not a long travel time not just the sort of stress of being in a hospital all these things are going to be beneficial and really that's where a lot of the clinical I AI is going to come in is helping to bring some of that hospital level expertise of your specialist consultant ophthalmologist and bring some of their expertise into your community practice clinic. Awesome well said.
Alright so we're going to kind of turn back I tell you it's going to cycle back to this and so if you as you're doing your research if you can narrow it down and go here the three four five things that academic institutions need to go here's kind of might to do list over the next year of how AI is going to impact learning and how AI is going to impact clinical decision making that the students need to learn what would be your short list of like here here's what you need to think about for the next one to two years like what's what's your bullet list. Yeah so the first ones unfortunately very boring regulation and governance and guidance you need to understand what your academic institution says about AI usage you need to understand what your professional regulatory body as I care professionals says about AI usage. So in the UK we have the general optical council they have a very broad statement that we should engage with technological advances and I health and health care delivery. So understanding what's being said in that space but then well that pretty generic.
What do we do with that. Why do we have something else. Yeah exactly so that's really what yeah what my PhD research has been diving into is trying to answer that question what what does that actually mean and what do we do about that. I think you know I mean you can even extend that past our local our our industry based socio political discussions into a much larger national and or global discussion I think governments.
Legislatures don't know what to do with AI and health care. I think there's a lot of vague and ambiguity on that and I think that's why that filters down as well well use it the best way possible with little to no guardrails. And I think it's some ways I'd be curious to see your thoughts on this. I think it's some ways we need to help establish those guardrails because I don't always think and for all the politicians who are listening I'm sorry if I offend you on this is that I don't think you know what's best because you're not in clinical care every day.
And so maybe the clinical care people need to help the academics help get on the same page and go we need to develop the guardrails because we can't count on politics to do that right I think everybody's looking to point the finger at everybody else and go what do you think. Yes, so I certainly wouldn't want to get in too much hot water talking about American politics not trying to set you up your your safe over there in Scotland it's good. But yes, so this is something that I do feel very strongly about that there needs to be more I care practitioners with their voice being heard in this area when it comes to regulation when it comes to policy. We're quite fortunate in the UK that we have a couple of very active professional bodies so the college of optometrist and the association of optometrist and they have a I task groups that have a wide range of people on them which includes experts academics includes data scientists but it also includes a lot of clinicians.
And so I think having that voice in there is really important and I think in the past has been up for me to maybe be a bit stronger at this than optometry of ensuring that the practitioners voices in there so I would like to see more primary care practitioners and optometrist getting involved in these sorts of conversations you know wherever that is in your local area. Cool. All right, so we got that was one bullet point any other bullet points you like hey if I have a short list of things to contemplate you know regulatory understanding what the rules are is always the place of the population in terms of you as an educator but also in terms of clenish you as a clinician so. The fun AI tool is going to be used in clinical care it has to be regulated by I assume the FDA and the USA and in the UK that's a great.
So understanding what does a tool need regulatory approval and then what is involved in obtaining that regulatory approval is really useful or sorry really paramount for clinicians and clinician educators to understand and to be able to pass that on to their students. Cool. Anything else Fiona yeah so slightly less boring than regulation I would say probably moving away from using the term AI quote unquote which obviously we're all guilty of we've been doing it this entire podcast episode but it's such a broad what we call umbrella terms so there's so many things that fall under the sun umbrella that you end up just hearing it all the time. And it almost becomes meaningless is just this big AI nebulous kind of blob so like same technology what is that yeah yeah exactly so I think getting into some of the sub types and that may be.
You know probably from an optometry perspective it doesn't necessarily mean getting into how it was made or you know how it was programmed but more like what does it do is it computer vision so it can take in some visual input like an image or a video and give you some kind of output of course that's very useful for retinal images or. Different images of the I you know are we talking about generative AI then let's say generative AI are we talking about AI as a medical device and let's say a AMD so being a little bit more specific with our language is probably going to be useful for for everyone to start moving on to. Yeah specificity right is you know I think that that's you know that's in she want I think that we you're right I don't even like the word AI artificial intelligence I like augmenting intelligence you know because that's really what it's going to do it's. Not making well sometimes it doesn't make stuff up but you know I that's great I look more specificity more designation of what what it is and how it's going to help my son who's an engineer says dad you don't.
Understand how a light switch works need to understand that if I click the light switch the light goes on it's it's causing effect right what's the cause of fact and I think that was well said in terms of how do we need to do that so any other bullet points before we finish up here. So I think from a clinician point of view if you're wanting to think about you know what should I maybe be learning because we're hearing about AI all the time and you start thinking I should maybe get myself ready but I don't really know how to do that yeah yeah I think having some understanding of what AI outputs are telling you and this doesn't mean from a tool specific level because if you have a specific. AI tool coming in the manufacturer will give you training it's not things like this tool tells you if the patient has glaucoma what I mean is that an AI tool is is giving you it cannot tell you the truth it can tell you a prediction of what it has determined is likely to be true and that is not the same thing is the truth well said well said and so the better the data that was used to train that tool so you know in terms of quality in terms of quality and it's not going to be a good thing for you to do it. So it's not going to be a good thing for you to do it in terms of quantity but also the more similar that training was to the scenario you now have in front of yourself the better that prediction is going to be but the they sort of accuracy that predictions not guaranteed and so.
So fantastic critical thinkers and so this is something that we need to be adding into our critical thinking when we're building up a clinical picture because we're not ever going to use one piece of information to create the clinical picture we're using lots and lots of information for the whole time the patient is sad in front of us or even before if we're looking to their old records and so just being able to understand. So we have an AI output and how do I start thinking about integrating this into the rest of my clinical picture is going to be really probably the most important thing for optometrists to know in my. Yeah Fiona I think I you know I was I was gazed these by how many notes I take and I literally have two pages of notes and like I never thought about that or that's a really good idea. You know so Fiona I first of all I want to say thank you for for joining us and for giving us we really haven't done anybody in real talk who's looked at it from quite your perspective and I think is a fascinating perspective about all the different ways that AI is going to change what we learn how.
We learn how we learn who teaches us how we provide care who we provide care to where we provide care. Fascinating discussion if you want to say thank you so much for your time I so enjoyed us having a conversation. You're very welcome I love chatting so happy happy to do it anytime. So thank you very much for listening to this episode of real talk with Dr.
Fueh and a Buckmaster please pay attention to this real talk as well as any of our others we release one every week at AI in I care. Come go to the podcast session you can listen to our real talk or any of our innovators podcast or we talk to some of the people developing the future so thank you everybody have a great week. You've been listening to real talk and I care you weekly podcast to keep you informed about AI technologies revolutionizing I care.