cat gbt or chat GPT 3 to be exact it has taken the World by storm if somehow you have not heard about chat GPT yet I can only say welcome back from your trip in the mountains and I could understand that you wouldn't know because it has grown so quickly it seems like it almost appeared overnight but actually open AI have been working on chat GPT for years now chat GPT 3 was made available to the public and in just three days it reached a million users 15 times faster than Instagram in just two months
it grew to a hundred million users which is four and a half times faster than Tick Tock grew so in this video I'm going to be talking about how chat GPT and AI in general as it continues to move forward and develop could completely change the way that you learn but how most of the people that I've talked to seem to be using it in ways where they're not really getting the maximum value out of it and if you're new to the channel I am Dr Justin sung I'm a former medical doctor now full-time learning
coaching consultant I've worked with tens of thousands of Learners over the last decade to teach them to learn more efficiently before we jump into the rest of it please make sure to give this video a like thumbs up to help with the algorithm you know how the YouTube Game Works let's get into it the first thing that I want to say is that there are a lot of videos and articles around chat GPT and AI in general but a lot of them seem to be looking at it in a very narrow specific way they're looking
at chat gbt rather than AI as a whole now some of you may know from a previous video that I've actually been studying artificial intelligence and the reason I was doing that is because for the last six months or so with my company I can study we've been working on a very substantial artificial intelligence and machine learning project in the learning science space in partnership with Google it's a quite a coincidence because actually at the time when I was going through that AI textbook and learning a little bit more about it just after that video
was filmed is when chat gbt sort of just blew up and appeared so it was very timely for me and when I saw it I was just absolutely blown away like everyone at what it could achieve and I used it like crazy I hit the limit basically every single day I've got a subscription to it now which I think at the time of filming this video you can't even buy it anymore even if you wanted to you I've been using chat GPT a lot like a like really really a lot so when I say that
I'm a fan uh that's definitely a case when I say that I know the limitations that's also the case I have trialled and experimented with it and a lot of ways to see how it can and cannot work in what ways if you use it it's actually more of a waste of time and makes things worse and I'm gonna be revealing some of those things in this video I'm not going to be talking about a procedural how to use chat GPT in terms of what you should literally be typing in there are some other videos
out there that are really good for this how teachers are able to use chat GPT to enhance and facilitate creation of their lesson plans and structures how you can create syllabuses or curriculum outlines or learning objectives or you can create practice questions how Learners are using them to look for sources or screen information and how you can generate not Knowledge from it by giving it very specific prompts and learning how to write the right prompts will get you a lot more out of the experience it's not just asking a question and see what it pumps
out you do have to have this dialogue with it to give it some feedback and to modify its parameters so that the next time it generates something for you it's a little bit closer to what you want and so there are lots of other videos and tutorials out there that do a great job at explaining this it's very important that you learn that so you know how to get the most out of the AI I'm going to be taking a perspective which is a little bit more macro it's a strategy yes you know how to
write a good prompt into it but what prompt in terms of the actual concept what you want out of it should you be trying to generate what is the type of work you want this AI to do for you which is going to be the most beneficial and what type of work do you not want it to do because it's probably a waste of time or detrimental that is the focus of this video and it might change the way that you think about how you can use artificial intelligence so a big problem that I have
seen and the advice that's floating around and some of the videos that are coming up around how to use chat GPT specifically for learning is that it is not looking at it from a point of view where it is long-term or something that is worthwhile for you to really invest your time in right now you don't want to spend the next few weeks changing the way that you study and your system to incorporate chat gpt3 right now only to find that when you have chat chipity 4 come out just changes things completely you don't want
to create a system only for it to be useful for like one or two months and then have to change it again so on and so forth until you know you're basically just changing your system the entire time now you want to be able to spend your time to create a strategy that is going to be able to leverage Ai No matter how that AI evolves and so if you're really speculative around what it can do now for you and these very specific use cases and then trying to think about well in the future maybe
a year from now it might be able to do this and therefore we try to build a system around that then you're never really going to be able to win this you're going to be constantly just trying to keep up with the technology you never want to be just playing the catch-up game you want to be staying ahead of it and in order to stay ahead of the game we have to understand a little bit more about what AI is doing for us and why it's really the same Trend that has been happening for decades
and nothing else has really changed except it's happening a little bit faster let me explain rather than just being Swept Away by whenever someone tells you about Chachi PT let's build some first principles understanding about what it's doing and what it really means for us and we can do this by just taking away some of the labels and terminology we don't have to call it AI we don't have to call it chat gbt let's just think about it in terms of what we need as Learners and how the system is structured around those needs humans
in society need knowledge and the ability to use that knowledge now way back in the day it was okay just to have a little bit of isolated knowledge we could have someone that knew a lot of these facts and that was enough that was valuable just knowing things and having information was enough back in the day now it's not so much the case now it's not enough just to know individual isolated facts no employer is ever going to ask that of you really now most of the time is how you're able to relate those facts
together and create more of a network and what that means is that we're able to see how individual facts and Concepts in that Network interact with each other and influence each other and therefore that means we can understand a little bit more context around it we can use that information and lots of different scenarios we can use it for problem solving and the better your network the more you understand these relationships and the more you're able to move between the ideas the more useful that knowledge is what's a good analogy for this let me ask
chair GPT chat GPT says think of knowledge like bricks it's not very useful just to have bricks which you can pile up and make just random pile of bricks but if you know how to stack those bricks in the right way you can build any type of structure and what's valuable is the ability to build the structure not just the fact that you have bricks was that a good analogy does it fit the concept properly is it accurate yes I think that's a pretty good analogy thanks Chachi PT and so for a long time now
in the research the type of knowledge structures aka the type of knowledge that exists in your memory that is actually valuable that allows you to perform at work and also in exams especially for the harder questions is called a higher order knowledge structure and something like solo taxonomy has actually defined this in Solo taxonomy we see that the later structures all involve integration relationships and extending and abstracting those relationships to new different contexts if someone has a knowledge structure that is of a higher order in Solo taxonomy then we would say that this person is
an expert at that subject and a characteristic feature of an expert is that they have this type of knowledge structure so how do you build this type of knowledge structure well the way that we build it is helped by Bloom's revised taxonomy and so Bloom's taxonomy talks about how the cognitive processes aka the way you think while you are learning changes the type of learning that you're getting from it so the higher order types of thinking especially analyze and evaluate which involves the comparing and contrasting finding relationships creating groups and not just creating groups and
relationships but prioritizing deciding which relationships which similarities are more important than the others which means that it's not just about finding similarities or finding differences but then saying okay here are all these differences and similarities which is better if we go back to our brick analogy it might be saying I could build a house by laying the bricks and arranging them this way or I could build it this way which way is best to build the structure that I want to build and you can see that at another level of Mastery it's even more expertise
than just someone that knows how to stack them in a good way in some cases we actually call this colloqueli wisdom the ability to understand context and conditionality of knowledge knowing when and how and why you should be applying knowledge in different ways and when it is not as effective that actually comes from higher order learning and higher order knowledge structures which you form through higher order thinking processes defined by blooms but it is much harder to do all of this comparing and contrasting and prioritizing it takes a lot more effort in your brain and
that brain power that we're using is called cognitive load and we know that some cognitive load is actually beneficial for us we don't want too little because it makes us a passive learner it's just a waste of time anyone can just read through it a thousand words per minute skimming through every single word and say I finished all these pages but absolutely learned nothing in their brain or you learn it but then you forget it a day later that means that the information was encoded to a low quality so what we want is we want
a high quality level of learning and the way that we do that is using these processes but those processes then take effort AKA it takes cognitive load now research on this has shown that Learners generally speaking tend to Veer away from this effort when something takes more effort we say oh it's too hard and then we stop doing it it's not because you're lazy well you may be lazy but it's in this example it's not because you're lazy but it's because we just want it that way we tend to say that if something is hard
we're not good at it and we try to Veer away from it now that's just a basic instinctual response obviously we can train ourselves out of that and that's also been shown in studies when we tell people that this load you're feeling is actually a good thing then people say okay well instead of running away from it they go towards it we see that a lot when it comes to working out we know that it takes some effort to lift the heavy weight but we know that weight is going to be good for developing the
muscle so we don't Veer away from it we lean in towards it the same thing is true for Learning and so this is something that we call desirable difficulty we don't want it to be too hard and too effortful and that's not going to be effective but we don't want it to be too easy either where we're just learning passively and so to summarize what we're saying is that what's useful in the real world and also for your exams especially at University especially at post-grad incredibly critical for PhD level and even in secondary school some
of the harder questions they are going to test you on relationships comparisons prioritization you're going to need higher order knowledge structures and to create those structures you need to engage in the higher order learning processes but those processes take effort which we tend to Veer away from but that if it is actually good for us and that's called desirable difficulty of learning and this entire process has many ways it can go right or many ways it can go wrong it isn't just step one step two step three there are dozens of steps and you have
to be able to monitor your ability to do each step and make Corrections if you notice that you are deviating from the path this is something that's called self-regulated learning it's the ability to monitor your learning and make corrections as necessary I am a self-regulated higher order learning coach not all of what I teach but that's a big core component of what I teach when I go and speak at conferences or do workshops and when I'm invited to speak usually the topic they want me to speak about is self-regulated higher order learning and this is
what my program trains people on so so in order to be able to make all those little Corrections and adjustments so that you can eventually do the right process so that you can eventually create the right knowledge structure so that you can eventually use the information and knowledge and retrieve it in a way that is actually useful and gets you good results even in school you need to have multiple sets of skills and if we break those down into four very very very general categories there are four key processes that you need to be able
to do to do any of this the first thing is that we need to know what to learn so that we're using our Learning Time strategically the second is that we need to find the right appropriate information to consume the third is that we need to know how to process that information in our brain once we have consumed it and that's really talking about doing those higher order processes that's a big big big big big topic I talk about this a lot in my other videos I train it in my program I'm not going to
be going into too much depth into that right now but that's the third step is you need to be able to actually do those cognitive processes in your brain to a high level and the fourth thing is that you need to be able to practice your ability to retrieve that information from those knowledge structures so just because you've created the structure doesn't mean you've become an expert you actually do need to train your ability to retrieve and use that knowledge as well so again it's knowing and identifying what to learn finding the right information processing
the information you consume and then retrieving and practicing your ability to retrieve that information if you can do those four processes as well you're generally speaking going to be a great learner now does that mean you need to do all of these processes to do well in your exam not necessarily because it actually depends on the way you're examined and also the level that you're examined at so for example especially in secondary school you're not getting tested too much at the higher orders it's still beneficial for your memory it's still more efficient and you'll still
do well like you know very very very well you'll be at the top of you know your game really by using these approaches but it doesn't mean that it's necessary at least not for everyone because the standard of assessment is lower it also means that you can get away with making more errors some people have higher levels of what's called Deep processing it's just your ability to think about information and understand it deeply if you have a higher level of deep processing that means that you can get away with having a subpar technique you can
have a technique that everyone else is using and then when you use it you can actually get more value out of it because that processing part is something that you're better at but if you're someone that doesn't have higher deep processing which a lot of people just call as just intelligence then you're not going to get the same benefit but the thing is that intelligence is not fixed it can be trained it's just a process that you are familiar with when you train yourself in a different process and then you become familiar with it you
are becoming smarter you're becoming more intelligent and so really it's just about what type of cognitive habits do you have if you've already got great cognitive habits from your childhood or whatever other experiences you can do well even with a bad technique problems in that technique will arise eventually it may arise in undergraduate or postgraduate studies or it might arise in the workplace but in reality it will always arise and I've seen this ubiquitously across the board senior managers at large companies talking about how their graduates are not performing very well because they don't know
how to think at that level University students are unable to do very well because they've never been exposed to that level postgraduate you know PhD candidates who are struggling to get through their PHD because they have never practiced thinking at that level in fact next week I'm actually going to Melbourne to teach some PhD students at Monash how to use their brain to think at their level because it is not something that is really explicitly taught which just bothers me a lot and is very frustrating for me but then on the other hand it means
that there's a lot of demand for my work so you know wins on lose some now here's finally where Ai and chat gbt all fall into things technology has allowed us to find the information which is step two much faster we don't have to sift through every single book anymore and we've been able to do this for a long time now Google search Wikipedia just in general the ability to use the internet has allowed us to find information extremely fast that bottleneck was you know like opened up a long time ago now Chachi GPT and
AI allows us to do this even faster because it depends on the question that we ask we can ask it a question and it'll just give us the answer however there are limitations chat gbt in the current stage at least at the time of recording this doesn't seem to inherently know what is true and what is false uh and those of you that have played around with it a lot will know what I'm talking about here but in its current state doesn't know what is factually true information and that's partly because of the way that
it's been structured and trained as a large language model but also because of the fact that it doesn't have connection to the internet now chat gpd4 506 I think that issue will probably be solved so really it's going to be good at finding you relevant accurate information and allowing you to then springboard that to find even more information however it still depends on the question you're asking it if you ask it a question that's not very relevant it's not going to give you an answer that's very relevant either it's going to tell you what you
have asked it but another limitation is that especially in its current state it is not at all very good at trying to create robust highly prioritized well-examined relationships it's if you give it too many parameters if you're too specific it just struggles it will still generate something but it will start actually violating some of the parameters that you told it to follow at a certain point it just doesn't seem to be able to generate that kind of knowledge and that knowledge is really what we're talking about is the stuff that's really at the higher orders
if you say do a prioritized comparison between this concept and this concept and tell me about which context it might be more important or less important in and what are some of the other relevant relationships that could form a comprehensive Network when I tested it against stuff that I have deep knowledge on what it gave me was quite superficial fairly generic not particularly unique it gives an answer that would allow you to have the level of knowledge of someone that is just slightly above average at that topic certainly nowhere near close to the level of
like a real expert at it and for those of you that are trying to do assignments for University level especially you're going to notice that this is not going to be able to get you those great results if it's just to pass it might help you but if you're trying to be excellent at something it's not there yet again in future versions of AI maybe this has been solved but from a AI like literally a machine learning perspective when we look at how we could potentially train that type of algorithm that's a very very difficult
thing to train partially because it's very hard to get what's called a performance measure or performance standard which is the thing that tells the AI that this was a good thing it generated rather than a bad thing there are like millions of different possible combinations of networks that someone could form in millions of different contexts to get enough data to say this is good and this is Banning give it performance measures is extremely extremely hard to do with the the current known methods of training AI so I feel that we're realistically probably still a little
bit away from it being able to do that now maybe I'm wrong and there might be this huge big Revolution and like a brand new method of training or algorithm that's developed that just bypasses all of these but I personally think it's unlikely and in my work with the AI project that I'm working on I've had the chance to talk to a lot of AI experts you know these people that do this day in day out for a living and they have also agreed that that's probably the thing that AI is going to find the
hardest to do and we're still a long way away from that so anyway technology just even the internet helps us with step two finding the information the other steps are now being helped with AI so we can use AI to help give us ideas about what we might want to start learning first we could say hey I want you to act as an expert on this subject and imagine that I am a complete beginner what do you think are the 20 most important concepts for me to start with and in which order and then write
a couple sentences next to it to describe what it is and also why it is so important to start with that in that order and then just by giving it that prompt now you know what to learn first that saves you a lot of time however a lot of people are not structuring their study in a way that is out of the order of the given curriculum that they're studying so they're not going to be able to use this technique because their study system inherently is linearly locked in the system that I teach and you'll
see this in many of my other videos I'm constantly talking about this not just in the program but also just in the free videos and I'm constantly pushing out there take control of the order in which you're learning information this is a great way to trigger you to do that you're set you're almost given a syllabus that are saying this is the order which you should learn things and then now you can have a list of concepts for you to start off and build a basic scaffold do your priming create an overall knowledge schema you
can use it for your pre-study you can use that for even during a lecture before a lecture afterwards for a review there are lots of ways that we can do that so step one is really really facilitated so there's a lot of time time saved there step two we've already talked about that's obviously there's a Time save there step three which is about processing information one of the ways that we can process information is by thinking about what are the relationships what is more or less important and how it could be important in real world
context as well as one other very useful strategy is to create analogies for different groups of information here's a very nuanced technique what you want is you want to spend your time with your brain doing the higher order thinking that's a desirable difficulty that's what we don't want to avoid but if we can get our brain to that point as quickly as possible that means that more of our learning time is spent effectively so when you noticed before what I did deliberately was I asked chat GPT to create an analogy for me so the act
of creating an analogy involves some higher order thinking however what's even more valuable is for me to take that analogy and consider how accurate it is so by doing that I'm taking my existing knowledge and I'm taking this analogy that's given to me and I'm looking for similarities and differences if there are too many differences this analogy is not suitable it's not going to be accurate or it's not going to be broad or comprehensive enough and so this is a way that you could use AI you can get AI to generate for you structures or
analogies or examples or metaphors to explain Concepts and then you could spend your time thinking about how accurate it is and whether you can extend that analogy into something that's even better or more comprehensive so this is a way where you're offloading the less useful task of creating the analogy in the first place and then allowing yourself to spend more time in the highest orders which is that evaluation phase another way is if you are creating a mind map or what I teach which is a chunk map and you're trying to think what are the
most important chunks and groups that I could form here to scaffold the entire topic you could give chat gbt some of your thoughts and ideas and you could use it to get other ideas and feedback you could give it certain relationships and ask it are there any relationships that are very important that I'm also missing to look for those areas so again we're allowing it to feed us the information but you can see it's heavily dependent on us knowing what to ask it and US feeding it information to facilitate the higher order learning the limitation
is here not in the AI the limitation is in our own brain most Learners because then not used to desirable difficulty and they Veer away from that most Learners are going to have study systems that do not have much higher order learning as part of it so if you don't have a system that has higher order learning integrated in it there's no way for you to leverage AI to facilitate it there's nothing to facilitate so what's important is to First have a method of learning that allows you to use higher order processes and then leverage
AI to do that faster and more easily I have been doing this extensively and when I figured out that this is a great way to use it my learning efficiency has gone up a lot I've been doing a lot of things not only managing this AI project I'm also writing a book chapter on self-regulated higher order learning published by Springer and in creating that book chapter I was thinking about all these different angles and different ideas cha gbt initially I wanted to just write the chapter for me I spent a couple hours trying to make
it do what I needed to do and then I just gave up because the output it was giving wasn't at the level of expertise that this book chapter would have required when I flipped my focus from just allowing it to be a sounding board and a gap checker for my ideas and my high flow suggestions then it became much faster and much better so if you're using AI to facilitate parts of your higher order learning that is a very good way that you can use it if you're using it just to tell you facts that
is not a very good way of using it because that isolated information is low value from a learning perspective as well as a real world perspective it's not valuable from a cognitive perspective and if lots of what you're doing is just trying to get it to tell new facts essentially the same thing is what you could do by reading a textbook but just you know if you would have read a textbook in a different order then that's not a very high yield way of using this very very powerful software and it probably means that your
actual method of studying is not very efficient to begin with so your efficiency is not going to become that much higher and that's important because remember at a certain point everyone is going to be using Ai and at that stage if you are using a strategy that is not effective even if you're using AI it's not going to be as effective as someone else that has a higher order strategy either using that plus AI you're always going to get beaten up you're going to be constantly playing catch up rather than staying ahead of the game
so what does it not help us to do well it doesn't help us to actually literally create the organization in our brain the act of creating the organizational structure and this type of high level knowledge structure in our brain has to happen in our brain the AI is not going to reach into your brain and do it for you creating those connections and and making your neurons move in different ways and firing the neurotransmitters in the way that it has to be fired that's something that your brain still will has to do and that's what
higher order learning creates that forces your brain to think in that way but if you aren't thinking that way in the first place well obviously AI is not going to be the one that makes you think in that way so it's not literally processing the information for us it's also not literally helping us with retrieving the information and this is one of the biggest limitations because we know that retrieval practice is so important that step four it's not helping us with step four very much we can use it to give us ideas on how to
retrieve that's very valuable but first of all it requires you to have a pretty good knowledge of different types of retrieval methods and interleaving and higher order thinking and learning in the first place so that you can give it the right prompt I have tried to just say act as a higher order self-regulated learning expert that focuses on giving interleaved retrieval advice and give me possible ways I can retrieve this subject I've tried that the recommendations that gives me some of them actually pretty good but a lot of them are not as helpful maybe that's
fixed in future variations but for now it's not something that AI is helping with very much but obviously even if it gives you the idea you're still the one that has to do the actual retrieval practice like you're the one that has to retrieve it from your brain because that's the entire point and so that is a very long-winded summary of AI and its effect on learning in the right versus the wrong way to learn it and I know it's probably not what you were expecting and it's probably much more complicated or do deep than
you anticipated but then again this is like a very long video so maybe you already anticipated and if you're a subscriber you already know that I make long videos so maybe you anticipate it again but anyway there are lots of different things to think about however you have to be thinking about it in this way there's literally no other option because things will continue progressing this way it has been progressing this way for decades now there has been a shift globally Asia has been very slow to catch up but even in Asia it's picking up
now higher order learning is becoming more valuable and the reason is because workplace requires this in the workplace companies businesses employers are going to be using AI for their processes of course they will because it is time and money efficient therefore you need to know how to use Ai and you need to know how to use your knowledge that AI does not fit higher order learning and higher order much structures these are likely to be things that remain valuable until the point where AI is completely overtaking every single thing about learning and human knowledge becomes
completely unnecessary up until that point higher order learning is going to be the last thing that goes if you are currently in school or university thinking about how you're going to be valuable effective as an employee or you know in the workplace later you have to think this way and if you do not think this way you will inevitably get left behind because AI is here to stay AI is growing Society will grow to integrate it the needs of workers are going to change higher education is already changing and the immediate kind of reaction to
it a lot of uh universities were saying oh well we're not going to examine we're not going to let you uh use technology but that's just a knee-jerk response that's just because they don't have enough time to change all their exams and their entire curriculum but in the coming years they're going to change their exams once they've had time to settle and process it they're going to change it and the secondary school is going to follow and maybe in some cases it already is following so so it's in your immediate and long-term benefit to think
about your learning as primarily an exercise of achieving higher order knowledge how AI can facilitate that if your current method does not facilitate higher knowledge that's a problem it's always been a problem it continues to be a problem the only difference is that now it's a problem that is approaching much faster because we are in an age of AI now you can't deny it and on the topic if you want to learn how to build a strategy that is more focused around building higher automatic structures then you can either binge on my other videos or
you can just join my training program where I teach you how to do it step by step that's another option as well I hope you've enjoyed this video if you have leave a like leave your thoughts down below there's a lot to talk about I actually had a list of all these different things that I wanted to cover and actually only covered three quarters of the list because this video is just getting way too long so there's I'm sure there's lots of different points and angles to talk about I'm not an AI expert I think
I know more about it than the average person because of all the learning that I did and the projects that I'm working on but I am a learning expert and putting those together I'm seeing a very very clear Trend so I would love to hear your perspective though anyway I'll take up enough of your time thanks for watching see you in the next one foreign
