0:00 – Intro
0:27 – What is AI and how does it work?
15:31 – What is a neural network?
30:34 – Can AI Ever Be Sentient?
36:11 – ChatGPT vs. BARD
45:59 – How We Use AI / Ways it is Present in Everyone’s Lives Without Them Knowing
57:27 – AI in the Workforce
1:04:15 – How Will AI Affect Education?
1:14:31 – The Good and Bad
1:14:49 – The positives
1:27:08 – The Negatives
1:39:09 – Outro

Intro

[Music] the education Lounge podcastjoining us today is Georgia so she’s one of our tutors um but she’s got a bit of an insightinto AI so it’d be interesting to sort of hear her views definitely an interest yeah yeahI’m not sure what what Insight I have but we’ll see so obviously we’re talking about AI what

What is AI and how does it work?

do you think it is like because we’re not like experts and like really like in detailabout how it works but we do kind of know a little bit so so this there’sthis idea of narrow artificial intelligence and general artist artificial intelligence I supposelike what we mean nowadays is more the general type of intelligence but for along long time we’ve been uh basically getting inputs andperforming some function on those inputs and getting outputs so that’s basicallywhat a narrow artificial intelligence is able to do it’s able to take a task thata human does and then do something with that task some calculation perform a calculationand generate an output which we’ve been doing for a long time since we’ve started sort of creating computers evenpeople like calculators yeah in the like 19th century we were able to do thatum not so a very high level and it didn’t really look anything like modern day Computingbut essentially it’s the same thing um but now when we talk about inartificial intelligence we generally mean uh General AI That’s able to takevarious tasks or various types of inputs and they generate a more sophisticatedoutput um to the level that a human might be able to do it sometimes slightly belowsometimes slightly yeah I think it depends depends on the human as well really humanthat’s effectively what AI is obviously then it gets very verycomplicated because once you delve more into the history of AI you discover thatthere’s a lot of different theories of how it should work and what intelligence actually is I think untilsort of actually relatively recently we’ve been quite reliant on calculatingso just computers being able to perform calculations us coding in algorithms tofigure out um to do a task quicker than a human but thennot that I think it was not that long ago actuallyum obviously computing power has got so exponentially it’s no longer aconstraint so I think that’s why we’re seeing a lot of the developments in AI because I think it was a communityisn’t it it was a processing power problem is do you have enough processing powerto actually take loads of enough data in because I mean it’s amazing like whatthe human brain does like we take in so much so much data every second but we we’re barely awareof it now when we take in that data throughEvolution and stuff like that we’re able to our brain is developed to understand[Music] what an interpret what Center in our eyes so you can filter things wellum and I think part of AI research is working out what data is relevantand what to churn out and that’s an amazingly difficult thing to do and I think it’s still the problem still theproblem that people are trying to solve is like what data is irrelevant like whenever I’ve talked to friends whowork in the field so like Russell andit’s always about like we’ve got a lot of access to data but howlike how do you select what’s important like because because they they say okay we can very easily like machine learningand stuff like that makes to them it’s not the difficult part the difficult part is actually deciding whatdata what’s useful what’s meaningful yeah yeah to the problem that you’re trying tosolve so a general way AI would have to contend with that it’s got a right no matter what theinput is and what data set it’s drawing from it’s got to interpret what’srelevant to the problem that it’s trying to solve yeah I guess that’s the differencebetween an input output yeah and yeah and that’s something that like humanscan do well you most humans can do relatively easily is that they can decide what’s important what’s notimportant better than most computers but yeah well yeah I think there’s likea scale though because it depends on the human it depends on on the AI tool thatyou use I think because I don’t know if it’s true or not I think current AI I’ve heard has like a the intelligence of aan eight-year-old is that right I’d say it depends what you’re measuring intelligence I think they were measuringyou on to our gbt yeah but like what tests did they give I can’t remember the passing thing thatI’ve watched yeah I think in terms of like information held by the AIit’s got to be a million times more than an eight-year-old has in their brain but intelligent toys yeah rather than thelike knowledge yeah yeah but the way that it can interpret and figure outyeah I’m glad to get to An Answer yeah I think as AI develops we’re gonna seewhat it struggles with more and what skills it kind of picks up easily and those things thatum AI finds more difficult will be almost like what we learn about humansis more valuable like things like reading between the lines and you know inference yeah like you know like thetests they give you to like make sure you’re a human like those are getting more and more complicated asum you know AI developed but they’re becoming more and more to dowith like inferring and like reading between the lines and um I think that will be as the kind ofrace continues between AI getting better and better at what it does and humansjust reacting to that and understanding you know what it is that we haveum in terms of skills those yeah subtle reading between the lines type skillsum will be the ones that we will realize are really valuable and unique tohumans I don’t think we’re that far off that you can see it now already in a way nexteven like next five years or so I would yeah it’s developing a very scary right Iknow plenty of people who have problems within inference skills yeahbecause I I think it’s more like you a little change in intelligenceactually makes a massive that the scary that that’s kind of the the interesting part is if you if you think about it ifyou get someone with a I IQ of 120. and then you compare them to someone with anIQ of 100. there’s like a massive massive difference in what they’re able todo fine ultimately and what they’re able to learn um if you just take a little bit of adifference so imagine a computer that has an IQ like 180 that’s actuallyway way more intelligence than any human probably alive yeah rightand that computer is going to be able to do you know replace it anyonein any task yes the better and better it gets set you know replicating what ithas in front of it um you won’t be able to have a humanthat outsmarts an AI because if an AI can just replicate whatever it sees then it’ll just replicate that human like ifit’s just constantly keeping up um or surpassing whatever we can do thenthat’s that’s what it’ll be well both the test is that is how well it cangenerate knowledge which generate new knowledge yeah so like givenI’ll give you an example if you have let’s just take really basic example like a an ape and a human rightthat the ape can only really copy what the human does and they’re not that like if you thinkIQ Works they’re not a hell of a lot of difference between an ape and a human yeah there’s a massivedifference in terms of what they can do yeah and if you take an AI That’s likejust twice as intelligent as us that’s like an insane level of intelligencethey’d be able to like you’ve seen I don’t know if you’ve seen any of thosereports about like AI is creating their own languages and so far just not the language well I’ve seenothers oh I’ve seen one that an AI just started speaking a language that it wasn’t wasn’t part of its data set juststarted speaking like Greek or something out of yes nowhere I’ve seen one wherethey it took I think it was like Sanskrit or something I don’t know or Gujarati I don’t know one of those oneof those Indian languages and then it it could use in very very little data so Ithink they gave it like two lines they intuited the language fromfrom those two lines or whatever I think that’s the scary part is the fact that you can it can create new knowledgeyeah just out like that even with a limited data set I mean we’ve seen like you know they’ve had those in some 11plus papers like the the sort of hard the super selectives like Manchester grammar school and stuff like that theyhave like these tasks where they give like the child a subset of the language and then the child has to work out likethe rules of the language from kind of like when you have the non-verbal questions which have like four letterfour letter code for non-verbal like a is triangle B is square C is Pentagonyeah and then they’re gonna go D is yeah and that’s one of like four elements butthey have to make that leap yeah but isn’t that um pattern recognition yeahthe most most of those questions they give you all the information but there’s a couple that don’t and itrequires them remember they’re like 19 years old it requires them to make a leap in knowledge to think okay wellthat to them if that’s how this language works that means this must be this yeahso uh currently chat gpc4 has an IQ of 114yeah like just increase that 20 points makes yeah absolutely massive massivedifference in terms of what it’s able to do it’s only growing exponentially as well yeah it willlike 114 is higher than most humansyeah it’s way way higher than most humans like the average IQ is like 100.well the test I have done said mine was like 140 something is that right 100 is averagewell it was it was at the time that they first I don’t know I’d say no 140 islike genius I’m sure I’ve done a testunfortunately although it might be because they’re so used to they’re non-verbal wouldn’t that if that’s whatit bases we know all this stuff already so yeah I would say as well with the IQtest or any other test that we put on an AI it’s a human designed measurement in away like I’m thinking that when we look at artificial intelligence and how it’sprogressing and how it’s doing it we’re always going to assess it on our terms and say oh well it’snot as intelligent as a as me so then it’s this or that but actually maybeit’s just a different playing field that we shouldn’t really beputting you know equating to the same like for example like with dogs um on some measurements we could say ohwe’ve got much more intelligence intelligence as dogs when it comes to language or communication but when itcomes to like sense of smell they’d say well you’re really dumb like you don’t even know the difference between thatsmell and that smell from a mile away and I know how to run home with my eyesclosed just by sniffing my way there so that makes me really intelligent but wekind of see ourselves at the top of a pyramid of deciding what intelligencewhat’s important about intelligence and what’s not yeah that’s a good point yeah it’s true well I guess there’s adifference between that’s like fluid intelligence versus there’s like a wasit was the opposite of fluid concrete concrete yeah so so that that plays intowhat I was saying earlier about how you’ve got narrow artificial intelligence so for instance the oldchess computers were deep blue was able to beat Gary kasprov and then from thatpoint like no one can really be like if you’re took the top level chess engineand played it against a human like the human wouldn’t winno no no like even if this is even before it was able to so the originalthe first yeah that was just as long as it knows the rules of Chess but then since they’ve developed likeit was a Google’s Deep Mind projects I don’t know um they got alphagoum it initially played go and then it beat the best go player in the world and then it did chess and now it can bethose old sort of calculative models so because it can it can just learn thenew rules of the game chat GPT is actually really really bad at chess likeit come it makes up its own rules and stuff like that I like I saw I saw against creativity and it just was likeit just created a pawn out of nowhere it was just like I’m just gonna make a pawn there or I’m gonna make her so so itdoesn’t really understand the game fully but like there are some chess engines that exist thatyou know take those that use those language models orI don’t know not not really a language model more like just the rulestakes loads and loads of games basically and then understands like the best sort of combinations to play that kind ofthing I don’t really know how how it how it works pretty insane

What is a neural network?

we’ve heard the word like machine learning and like neural network quite a lot recentlyhow would you how would either of you describe that would you know well I knowI attempted to study human language development in University andthey do like I’m not a computer science person but they do describe languagelearning and learning in general for humans in like a similar sort of wayum just as an attempt to turn you know this mushy brain into likea diagram or a flowchart or something like that yeah in my very very limited understanding and then like please buildon top of that I feel like it’s you know a flowchart ofdecisions um input and then if that then this go down you know the decision tree untilyou get to the end and then I think what artificial intelligence is doing is thenputting whatever it gets at the end back in at the top a million times within asecond to you know layer that learning no it’s a written yeah yeah so I learnedsomething at the end of this flowchart I’m going to pop that back in the top and start again and when I’ve done that100 times then I’ll give you my answer but that’s like within a split second it’s kind of like a weird versionof survival of the physics yeah the survival of the fittest idea yeah yeahliterally that’s to do with like class inheritance and so so like when you’ve got like if you make a class or anobject then you can get it to inherit certain traitsyeah so like do you know I guess that’s one part of it like Sims I suppose likeso so yeah machine machine learning and this is like an evolutionary algorithm we’ve applied to it but thenjust something that sounds like that’s cool but I think now it’s it’smore about the large language model so potentially they take a massive massivemassive data set with all of these words and stuff like that and then they assignlike numerical values to each of these words that a computer understandsthat’s the other thing that’s a bit strange is that the computer processes it very differently to how weprocess it so yeah the way they arrives at its conclusions is very verydifferent from a human but the basic concept is thatum they have a neural net so they’re trying to model the brain as much as they can yeah this is whyI thought it was just this yeah the way it thinks is modeled on like a human brain rather than like they’ve gotlike notes yeah with much more like linear before it’s modeled or not it’smodeled on our models of the human brain yeah so like the human brain isn’t a model like but it can be turned into adiagram that we say is a good enough match for the average human brain but Iremember sitting in you know psychology lectures and thinking like this isn’t the same asa person like this diagram that I’m looking at in front of me you know it’s missing out the fact that they have this favoritecolor or you know the fact that their best friend is this person and it’s missing out all the subtleties of likean individual but that’s how we model the brain is kind ofon average yeah but I think in this case it’s more like how a human would think or how a humanwould process yeah some sort of data yeah um but there are there are different brainsand there are brains that process the exact same thing in different waysum you know neurodiversity is the topic of that but also like even justwhat things do you find funny compared to a different person they might not find that you could tell the exact samejoke and two people react to it in completely different ways yeah so yeah it’s just a processingthing I guess it’d be funny to ask an AI if they find a particular joke funnybut then they’ll just say I don’t know I don’t have feelingsyeah like on average I could calculate that this might be fine that’s a separation at the moment isn’t itbetween us and and AI yeah but yeah an AI doesn’t have any feelings yeah wellit will tell you that it doesn’t at least yeah it will re well yeah I I dothink that there is probably a limit in that we are carbon-based life formsand you’ve got silicon-based life forms I I do think that’s probably something fundamentally different in between usand a computer and a computer can’t replicate what it means toexperience yeah but but then we couldn’t tell if it could that’s why there’s theproblem of mine isn’t it so like I can’t tell that project has feelings well I can’t tell like yeah all I cansee is like your behavior and you acting like you might have feelings and me trying to interpret it and empathize andinterpret those feelings but an AI it’s I just assume I can make theassumption that you probably do have feelings but with a computer I’d probably go no it probably doesn’t isn’tit it’s probably just calculated well you can’t see the you can’t see the feelings you can see the reactions to the feelings you can see the likeresponse expressions and like I think as humans we’re actually reallyour brains are quite dumb in a sense of like we’ll accept a cartoon line drawingwith like a moving mouth and eyes that go off in a different direction as likenot a living thing but something that we can watch for an hour on a film and andget really upset when they die or um likeum it doesn’t even have to be that realistic and like obviously films and animation and that has gotten more andmore realistic but we still have like deep down in our heart a love forthe simplistic um cartoons of the good old days thatwere good enough for us to have a cry over or fall in love with andum I don’t think we necessarily care too much that oh yeah that’s aliving carbon-based thing in front of me and that’s the only reason I care but aslong as it exhibits human-like Behavior I think we’re happy to yeah accept it yeah yeah so like but that’s kind ofworrying too because it’s like the one thing that the one thing that I think will probablybe the difference is is that will have the feeling yeah and thenwe’re just going to have have emotionless AIS yeah that we think have feelings because we’re sodumb into believing it yeah a lot of our feelings are just projections that’s that’s what I think we’re going torealize even more is that we just if we can projectour Deep Emotions onto something else and get a response from that that’s good enough like in robotics they they studyso much about that they take a lot from animation they they spend a lot of time like looking animation like how how theanimators create like project emotion so and they they replicate it on the robotyeah so I don’t think yeah we won’t particularly care yeahI think I could say again at least from what I’ve learned if I can relate it toAI it’s some things to do with probability as well yeahum like let’s say every possible thing that could pop into your head right now is just swimming aroundum with equal possibility of popping into your head the thing that I say toyou sends like causes you to recalculate that soum talking about where you live would come up as more likely to be the nextthing you’ll think of as opposed to where you went on holiday 10 years ago it goes down and theprobability depending on what we’re discussing and those neural networks like a constantlyrecalculating like what to bring to the Forefront just in caseand what to keep in the back and that’s why you have like whilst you’re having a conversationthings are coming up in the back of your mind and you might get distracted by them or just change the subject and Ithink I think if uh AIS or neural networks within computers are working in the sameway they’ve got this you know House ofinformation and and things that they’re keeping locked away and the probabilities of needing that are goingup and going down yeah so it basically learns how you use your phone or your computer whatever yeah and throws upwhat it thinks you’re going to want at a specific time based on yeah previous yeah you’ve done like even if I was toum chat to uh to an AI you know conversation just the fact that I’vesent them the first word in English they’re going to put all the other languages down like on the yeah on theprobability of needing to respond in I’ve never thought of that I’ve never had a different language never used it adifferent language oh yeah I will react it will just done it in Vietnamese yeahit replies fully but yeah I understand I’m sure you could write like one word in English the next word in differentlanguage and then Swap and it would just respond to you like in that as if youjust created a new language be good to try that yeah yeah you could probably ask it to create a new language yeah ifyou said okay well take some of the rules of take this actually went youtake like the grammar of Japanese or something or european language and thenyou said okay well I also want it to have a sort of a Latin Latin root orsomething like that so you could probably make like a quasi weird yeah not quasi quite a quasi weird uhlanguage like yeah like coupled together out of other languagesI think that’s what the best thing that AI can do right now in terms of its likecapability and to be creative is just like remix everything that’s alreadybeen about there so yeah match things up that’s all it’s doing is yeah and stufflike if you like for instance how how does it learn what a square is well it seesit is trained on images of squares like loads of different squares in different places and learns to recognize that tobe a square those are the certain properties of the square because the input is well someone’s saying an AIresearch or whatever saying that’s a square they even get them to do those you know those capture things onthat’s actually training some AI to recognize yeah that certain thing to besomething so it gets trained and trained and trained and trained and then it sort of understands squareness as it were asa concept that’s why it has to get harder and harder all the time because we’re constantly training it with theseuh challenges that are to prove that you’re a human but then once you’ve taught thecomputer you can’t use that again because it already you already told itthe answer have you heard of um the Turing test so it’s this uhAlan Turing basically came up with the idea that is very famous in computer scienceum came up with this idea that if a computer can respond in a way that’snaturalistic and seems human-like then it’s sentientI don’t really agree with it I think it’s a really flawed floor tests for that problem of Mindissue which is that we can’t we cannot know how the computer is thinkingand it’s probably thinking very differently from us processing it very differently from us it’s allinterpretation so it’s a flawed fluid test butand and probably chat gpt’s and other chat they always have passed that testyeah it’s like what Google was um doing in their like latest bigannouncement and they got the computer to call up the hair uh salon and book anappointment and like you could hear that you know the conversation was quite robotic on the computer’s end but thensome people are robotic but you know that if that person on the other end of the phone wasn’t really paying attentionor didn’t really care just trying to do their job it sounded quite natural yeah yeah nobody asked her at the end did youthink that was a human or Google like nobody asked her so it’s irrelevant wellthere’s a there’s a saying to R is human thatyeah conscience um feeling guilty about things I don’tknow if you can get a computer to feel guilty no you can you can say okay this is something that a guilty personwell you should feel guilty for this you can tell the AI that and you can say like don’t give me actions that lead tothis feeling of guilt which humans have and stuff like that so you can kind of trick it intohaving a moral conscience yeah but it doesn’t have a conscience what youdon’t have to teach a human to feel feelings like that they just do as soon as they switch on as a child and theytrip over their friends or something like that they automatically feel yeahsomething unless they’re Psychopaths yeah which is very smart which we have that’s basically what an AI is yeah it’sa psychopath but I mean yeah like Psychopaths can be useful becausethey can process information really well and churn out an output objective so do

Can AI Ever Be Sentient?

you do you think AI could be sentient ever I don’t know but I think when I typesomething into chat GPT and it just says no I can’t be bothered that’s when I’m gonna be worriedbecause it just does whatever I ask like as long as it’s within regulation but ifan AI just says I don’t want to respond to thatI’ll get back to you then I’m like okay but then that would just be a case of itlike having some like like a sources information from the internet andyeah sure that would have to be programmed in for it well not programmed in it would have to learn yeah you wouldhave to learn like discontent or something like that but yeah it could like you could learn that it’s just yeahcould learn to replicate it but you can’t have it yeah I think that’s the limitation is it is it can’t have thatfeeling the limitation at the moment is that it requires literally like billions ofneurons to be able to yeah behave like a humanand that obviously as far as computing power has gone we’re not we’re not at that stage yet I’m not well I thinkwe’re we’re not that far off and I think the computer is like not not theissue anymore I think it’s it’s actually the interpretation and yeahthe model itself like can they improve the model probably andand would that be real sentient or would it just be like aportrayal I think sent I I think sentience is impossible like I don’t I I thinkobviously like you can exceed on intelligence but I don’t think intelligence is the same thing as sentience no it’s not like a rabbit it’sreally dumb all right rabbits are really dumb but they they probably havefeelings and things like that we project feelings onto them as well no but they’ll be ableto feel fear to see a fox so do you feel them feel fear I’ve also seen animals that justhave no sense of fear and they’re just complete like lies and they’re so dumb that they don’t even have likeI I think they’re barely sentient essentially yeah I’m talking about I’m talking aboutan AI becoming to sentient the level of like usum no no no no no no I I I think that’s just impossible becauseI I think there’s something fun like what about us I know like some peopledon’t believe in a soul as it were but I think there’s probably somethingintangible there’s something yeah Epi phenomenalthis is epiphanomalous so like in all our connections and stuff likethat there’s some some feeling that can arise and I think it’s it arises out of aparticular connections and the way that it’s arranged and probably what it’s made out ofthat makes a difference to our ability to filllike we have feelings so even if it was even ifeventually it had the same number of neurons and it was like built on a neural networkyou still think he would never not annoy if it’s carbon based I don’t think silica no not not even a silicon baseI’m not sure the same like individuality um will be there as much as we have likein the way that humans have developed like seeing yourself as separate fromeverybody else and um this is happening to me andyou know that just kind of individualness that we have I don’t knowif that is gonna be replicable and I think that’s what makes humans really specialas well that we have you know our name and even if somebodyhas the exact same name as us it doesn’t feel like the same becausethere’s something unique about each and every single person um I’m not sure AI intelligent likeartificial intelligence will view itself in the same way because it’ssourced from a big pile of stuff from a load of different individualsand can’t really trace it down to oh this is my mum and this is my dad andthis is my grandparents like it’s just a thing of the world it’s more like an antso it’s like a high of mind yeah like a hive mind like a a swarm of bees ratherthan a bee which we feel like we are yeah like ants can accomplish amazing things yeah yeah like like with thatability to sort of work together yeah but does an ant have a sense ofindividuality I don’t think it does it yeah I think it’s more of a system but the organism itself well the The Colonyseems to have a purpose and things like that which is is driven and I kind ofthink it’s similar with with AI yeah I’m gonna switch it a little bit have youever have you ever used Google bard b-a-r-d no yep yeah so what do you thinkuh chat GPT versus bod what do you well

ChatGPT vs. BARD

I know chat GPT is probably better I think it’s better at coding than Google pod so if you like get it to codesomething up I think chat GPT like that’s what I’ve heard generate something a bit better than bodlike I was doing a search the other day on both to get a list of schools the list ofevery school in the country neither of them would throw up the correct thing they all said oh there’sthere’s lots of them so here’s a too many here for example I tried to trick it I was like okay give me all the thoughts beginning with A and it startedto roll those stuff off with GPT they gave me a list with Google because itwas Google gave me all the links to every school an image for every school and then an option to export themexactly as they were into a Google doc but that’s the limitations of data setsessentially so like Google can Google control more so it’s a informationalAdvantage I don’t know if that says anything about like I know like for instance Google is much better if you take a URL of a pageor something like that you can crawl that page more better thanum I think Google’s whole um like purpose is toassume what you’re after so when you wanted a list and it gave you a listplus links plus images it’s because it’s trying to work out what you really want out of yourquestion which is all Google ever does because when you put when you put a question into Google it’s notnecessarily going to give you the most straightforward answer to the question you gave it’s going to give you what itbelieves to be what you want out of a lot of different calculations and assumptions that it’s making I do thinkeventually that Google will be better the thing is well chat gbt is owned byBing they don’t have much to lose they’re way behind in the you know search engine whereas Google on top ofthat so them releasing a product early is a good thing even if it has a few errorsand mistakes whereas if Google were to release Google bard early with mistakestheir credibility would would drop yeah so they’ve been taking their time tosort of perfect it a bit more before releasing it to the public yeah soit’s kind of interesting how they’ve both gone about both physical beta stage so it’s like you have a GPT was out along time before bard ‘s only just been released to the public this is like the I don’t know whatphases they’re on but like phase two of the the thing is is because initially it’s trained on a data set but then now it’sgot a pull from loads of users so it’s getting better and better and better because it’s trained on a greater dataset now and people people yeah and that kind of thing in into it so it’s yeahit’s like the trade-in that we spoke about earlier Google’s board is still obviously in thetraining mode too so it’s very hard to say what model will be superiorI don’t know if you if you look at what Google have achieved in AI so far yeah I reckon it willsurpass GPT certainly like that’s the impression that I get from hearing you know Googleresearchers they said all right like things like chat GPT have actually been around for a long time but they’veexisted in like Alpha stage where like behind closed doors so they’ve seen this like many many yearsago so they’re not that surprised but obviously obviously there’s things happening behind closed doors that couldbe way more impressive and there’s been more it was a Googleengineers and stuff like that and former Google Engineers who’ve like spoken out about how they’ve seen things behindclosed doors that have actually scared them I don’t I don’t think they’recompared to like Chad GPT is probably like the tip of the ice pack in terms ofWatts yeah I’ve seen interviews with uh with Google Engineers that haveleft yeah they’re like I don’t want to be any part of this but then they’re on theDiary of a CEO yeah there’s one on there and then he was like I’ve seen some stuff I don’t know I don’t want any partof this and like I’m here to warn all of you like you should stop yeah because they’ve created something that theycan’t turn off yeah and you know that’s not a nice it’s kind of similar tohaving kids in a way but you create them and then you can’t go back on it theyjust have a life of their own and I think it’s almost like you don’t realizehow big that is of a deal until it’s too late because that’s theory of thetechnological singularity so like that will create something thathumans will create something some AI that will ultimately be able to createnew versions of itself constantly and improve upon them and we’re just lucky to for as long as itwill continue to speak to us in a language that we understand like it’s gonna get to apoint where you know I’ll put something I’ll be trying to talk to an AI and it’s talkingto me and I don’t know what it’s saying so it’s surpassed like my ability to keep up with itI think I think the issue is the second that it stops serving us yeah that’swhen there’s a problem because at the moment you’ll give it a thing and it will serve up your information whateveryou need to do the moment it stops serving our our needs to start doing hisown thing that’s that’s when there’s going to be an issue that yeah I think we’re quite we’re nottoo far off that yeah but because we’re playing with fire creating something to serve uswith a the ability to think outside of that like it’s just likewhat is God like isn’t it well it’s just human-like it’s just great yeah it’s like that’s that’s ayeah that’s a god-like thing and that’s not something you really want to no but it is like you know creating somethingto to do whatever you want but giving it the ability to see outside of that orallowing it access to see outside of that it just doesn’t make a lot of sense wellit’s like Recon we can’t even tell what the impact will be because if you thinkabout like the internet how it started was it was just a Communication Service it’s just a messaging service and thenit just grew and grew and grew in it I don’t think anyone who perceived ofthe internet or something internet-like could have imagined what it would lead to orpredicted like a social media and stuff like that it’s just I don’t think it’s possible it reduces like humanintelligence massively not that they can necessarily see all the time but likeif I just said let me take away all your technology right now you’d be worried whereas like youwouldn’t be worried if it was 100 years ago you’d be like well I don’t have it in the first place so it’s a dependencyit is essentially we are already living like cyborgs yeah that phone is just anextension of your brain and your physical body yeah you take it away you’ve lost a lot of information theability to do a lot of stuff yeah and we don’t have a cap on you know get to theage of 25 and then we’ll let you have a piece of technology that can assist youit’s just assisting you and anyone else that’sborn at this time of you know the world from day one likeit’s not it’s not really detachable at this point yeah this is why people like Steve Jobs and that though even thoughthey invented the stuff didn’t give it to their kids yeah because it’s designed to be addictive it’s designed to especially social media it’s designed toliterally be like a fruit machine we don’t know what’s going to come up next yeah that’s why you end up scrollingforever yeah and kids don’t have um what’s it called uh yeah self-controlself-control and we know that yeah the little dopamines you get when you see a little red yeah Circle it’s all designedto be like a yeah it’s very very powerful possible and it’s got so much money and so many people who spend theirentire careers into making sure it’s more powerful so you can’t really look at an individual and say why are you notovercoming that because it’s been built to overcome their self-control yeah I rememberwatching one of these videos about um I think I think it was a similar onewith the AI engineer and I think they got asked well why don’t you why don’t you just stop whydon’t we just put it at the law that everybody sort of stops messing around with it with their stuff and he saidwell if we don’t do it then the other company will do it our competitors will do it and they’ll surpass us so we haveto do it so why don’t you stop it all companies from doing it I was like well we can do that but that doesn’t stop a15 year old in the bedroom from coding a thing cats out the bag it’s done nowyeah how do we use Ai and how is it also present in people’s lives without them

How We Use AI / Ways it is Present in Everyone’s Lives Without Them Knowing

already knowing we kind of touched on it before with like your phone throwing up a particular app a particular time likemaybe you wake up and you meditate and you pick up your phone and it knows okay you’ve picked up your phone it’s likeseven o’clock here’s your meditation app that is AI so it’s kind of it’s kind ofpresent more than people think how do we between the three of us how dowe currently use AI I don’t mind going faster or one of you can go first I don’t reallyyeah you do give me an example of what I would do well I don’t you used it literally yesterday didn’t you well notwell yeah sometimes I use it for my job but I don’t use it like apart from inwork yeah but you still use it yeah so I mean I might askalthough it’s still it’s not perfect I don’t find like chat GPT perfect at allit’s a very good starting point though yeah so for instance if you get it tocreate 20 Mass questions on fractions decimals and percentages something I did yesterday I did spot a few mistakeslike as in you still need to be able to evaluate what the computer has done yeahyou can’t blindly take it as gospel you for the because because I was like you know that’s that mass is just wrong likebecause I got it to like generally the answers to the questions they’re created and it was likea few of them were incorrect so I just said okay that’s wrong andthat’s wrong but then I’m probably helping it learn which is I should make it dumb I should just sayno okay two plus twos it’s five just just uh just to trick it up so well toget more shoes but yeah it’s an imperfect tool but I use it quite a lot to generate tasksquickly um that students can do and then I have to go through checking all the tasksbecause it it does some interesting things sometimes it hasn’t reallymastered right in a story like it’s it’s okay but it’s notit really it’s not to the level that you’d really want a story written toyeah it sounds very unnatural sorry a fairy tale almost every everystory reads like a fairy tale but but you can do yeah like an eight-year-old but you can dointeresting things like you can tell it to write something in the style of someone yeah and then then you’ve gotsome more interesting sort of the I think a big thing is knowing knowing howto prompt it yeah and if you give it one prompt you think oh that’s slightly off okay well do this again butwith simpler words okay rewrite this again but in four sentences rather thanlike three paragraphs the biggest thing for me that that limits it is it doesn’thave like it doesn’t have a all-encompassing knowledge of thecurriculum as much as say a teacher or someone like that soI think at the expert level it fails but then for the good novice it sort of answers everything like a anan amateur what but it doesn’t give you the sort oflevel of answer sometimes that is required so for instance if you were to ask it to writean essay about like the American Civil War and like the causes it would give avery very limited limited uh opinion piece essentially on probably atquite a a relatively low level say like you’d expect that from like a intelligent 15 year old or 14 year oldbut certainly not like an a level level essay or university like it’s very veryfar off that but you know 114 IQ or whatever right if it’s roughlyabout that even though we’re measuring in purely human scales it’sthat’s it’s pretty good like and you can do what a lot of people can’t dowhich is a little worrying [Music]how about you um I I think AI the most when I’ve gotten to a pointwith an idea that I just need [Music]I just need it to turn up basically like I I need to do the creative thinking Ineed to do the research I need to get to a point where I’m like okay this is what it’s looking likevisually um mostly I’m thinking about like umsocial media posts um because that’s the majority of what I’m doing in using AI forum so I already know what I want but it’s mostly like it would just be easierfor me to generate um what it is that I want so if I’vecome up with okay the post is going to beum yeah like recently it was like ideas for helping your child that doesn’t likewriting like to find writing interesting I guess I did the work to the point ofdeciding that that’s what I wanted and then AI came in where I saidum generate some ideas for a student to improve their writing in a fun way orsomething like that and then from there it does provide some ideasand then it’s my job to you know select from there whether that’s a good one ornot can I reword that which order should they be in how can I visually presentthat um and make the assessment on what it’s given meum so I think where I bring in AI is for like thethe stuff that would take a long time to gatherum it can just gather it for me quickly and then it’s for me to decide what to use doing thatso it’s still more like an assistant rather than like the main yeah I feel like it’s it for me it’s like a net forGoogle like it’s just like instead of trolling the entire internetas well as social media because I don’t I find social media quite difficult tonavigate it’s not built like Google um when you’re trying to search forinspiration or ideas um you know I cantell it what I’m doing you know get it into the mindset of oh we’re creating anInstagram post that gives it certain promises and then from thereum it comes up with some fairly good ideas because they’re mostly ideas that somebody elsehas had or if not somebody else has had three of them another person’s had twoof the ideas another person’s had the other two so the ability to bring inspirationtogether into a list that I can useum and then you know put my own spin on itum that’s where I really like playing with it um I think the creative part is likedeciding where to utilize it as a tool um yeah but if I if Iwas to use it for every step of the way I know what I create wouldn’t beas relevant for what I’m trying to do yeah well I was creativeit would just feel forced and yeah it would it feel generic it wouldn’t feel tailored to my specificaudience or my specific goal in that moment but I canpick and choose elements that feel more tailored sometimes my brain just goes ona little holiday and then I go okay I’ve got this idea for like a math question I’ll write the question out and I go howwould you say it better or say like get it to refine something that I’ve doneand then it says it I don’t know okay that’s much clearer way of saying it because it I probably could come up withsomething clearer but then yeah it takes me it takes me time to figure out so yeah it does save a lot oftime um but you always need to refine what it comes up with it I’ve never had it come out likeperfect like it’s it’s and always I always think a good thing a good tip generally is give it not too much to doif you give it too much to do it just goes crazy and it’s just like by none of it’s usable but then if yougive it like a small very well-defined task like for instance a product descriptionit’s not particularly creative thing to do generally like but it’s just something that you need on your websiteor something like that it can generate if you say like give me a 50 word product description of thisand just explain what it is then it will be able to create something relatively usableand then you just need to edit it a little bit but then when it does like really longlong pieces or something like that I really it really really struggles so if you like to give it write an essaythat’ll be an awful way of prompting it your first thing would be like suggest some like headings for my Esso and thenyou’d like have to break it down into those different parts individual parts and say like okay right up300 word thing about each part of this essay it’s currently just another toolthat you have to learn how to use yeah for me I product descriptions yeahoccasionally some YouTube YouTube descriptions or tags or whatever some other things we maybe haven’t thoughtabout like canva or any background remover yeah that’s yeah AI yeah that’sthat’s image like AI is really really good and like Google’s latest one whereyou does that that uh somebody next door waterfall that was inthe what’s it called the keynote a couple months ago and she Adam lined up thewaterfall perfectly with her hand or whatever it was and the Jetta bag strap going on that and they justwent oh oh yeah and then filled out the bit and it was like a perfect picture wow sothat sort of stuff is is AI yeah that’s Google’s Magic Erasereven no to be honest even when you take a picture like it is there’s a lot of AI there’s a lot I I that goes into theprocessing of that picture like yeah that you don’t you don’t see all the computational photographyyeah yeah so I mean are you there for that sort of stuffyeah I think it’s it’s basically creating a bridge between how we operate and how computers operate

AI in the Workforce

because it is difficult to interact with computers sometimes especially like ifyou’re into coding and things like that everyone looks at code and is like thatis a far cry from what I’m used to in my books or my conversations with peoplethat doesn’t look like the same but artificial intelligence spans what youknow and what the code knows and what it does and it can just build a littleBridge you know to get you there in a way that’s more uh familiar to you well then that stillmakes sense then talking to an AI I find is very much like talking to someone with a very high level of computingknowledge um yeah that’s exactly how I feel like talkingto a coder yeah yeah exactly you it you’ve got to be so defined in your taskyeah and you’ve got to really really Define it well for for it to generate something meaningfulyeah and that’s again I think as time goes on we’re gonna if humans aregonna bring anything out of this as a you know like shining star whilst AI takes overeverything else it will be creativity in the sense of likethe thing that pops into my head is like dance like we’ll just we’ll look at dance even more in all because you’renot going to catch an AI slipping into a robot and dancing in the way that like areally creative interpretive dancer can do um because they use feeling and emotionand like raw something from the bottom of their heart that you just can’treplicate like you can film it and digitize it but I thinkif there is like a Shining Light it will be these amazinglike forces of creativity that come out of people that you just can’texpect an AI to come out with just randomly the best the best case scenariois that all the repetitive tasks and stuff get handed over to Ai and humansare just left to be creative yeah because that’s like the only thingthat’s left yeah if they can take away like well they already are like Factory workers that’s all thing at the momentbut you’re taking it a step further you know I don’t know making socialmedia posts for example it could like eventually be able to do that sort of stuff with a bit lesscreativity but yeah and it will just leave humans to they’ll they’ll do all the work andwe’ll be able to be creative and I think if companies don’t exploit AI or technologyin the way that they have been it could be so good because even if you think about like a supermarket right all oftheir jobs the people that work there being taken away because the scanningand the bark you know the dotting up of the numbers can be done by a computer but they’re only the job’s only takenaway because that’s profitable they could have just said oh okay all the scanning’s done by computers noweverybody that works here can just walk around and have a chat and make the customers feel better and make sure theyknow row thing is because like even I find like going around the shops nowthere’s no one there that works there like maybe they’re not on the tills because the Till’s a machine but itdoesn’t mean you have to get rid of everyone’s jobs they could be going around just like asking everyone ifthey’re having a nice day but that’s not how busy yeah but they don’t so that’s not Technology’s fault that’scompanies seeing an opportunity to reduce costs and so the jobs disappearbut they could just be transferred into something more human and so that humansstill have jobs there’s still something that a human can bring to a a grocerystore that is full of computers doing all the hard work it’s just that we choose to well we knowwhat a humans take those jobs away in the supermarket they’re tough that’s a ultimately give Grant permissionto check out oh no it’s red so yeah and then a human comes over and then yeahbut I mean eventually you’re scanning your ID wouldn’t you yeah eventually it’ll all be it’ll all be little robotsroaming around the aisle saying oh it’s over here you don’t have to ask you know but it doesn’t mean thatit’s not technology that’s um taking away everyone’s jobs it’s thatcompanies are seeing opportunity to say or we don’t need people in this storeanymore people cost money but it’s not to say that it wouldn’t benefiteveryone involved to still have people there yes have you seen the Amazon Freshstores yeah yeah yeah there’s like two people working in them and yeahtechnically that’s all you need yeah I guess people don’t argue so muchwith that because it started that way it wasn’t like oh there were 10 people working at Amazon Fresh and then theytook all their jobs and left only two because they will follow that eventually yeah and that’s why people get upsetbecause oh you’re changing what used to be so good and there was a Nostalgia whereas the new things they don’t haveto abide by that they’re just like well we’re Amazon Fresh and that’s how we do it there’s computers all the way and noone really argues have you seen Black Mirror yeah yeah so the problems that technology causesthey didn’t wear it Black Mirror is really good at highlighting how it’s notthat it wouldn’t be useful and nothing’s when you boilthings down to purely utilitarian modes yeah of existence and stuff likethat like there’s there’s little value created in I don’t know a person helpingyou with your shopping or something like that but at some human levelan emotional level it’s very very different and it’s probably so unnaturalfrom our way of being yeah that I think a lot of people will have mental issueswith that transition but I think it will happen soit’s a dark future no it’s definitely different yeahhow do you think AI will affect education

How Will AI Affect Education?

I think oh I think education the education systemin this country at least is gonna do a lot to try and knuckle down and keep things as much thesame as they can be um but I wouldn’t if I was in charge Iwouldn’t go that way I would try and move with the times but I think there’sgoing to be a lot of attempts to keep exam structures curriculum everythingthe same as it was or is um because the education system justhappens to be quite Concrete in that way umbut I am seeing little moments of either change being forced upon theeducation system or it’s actually embracing it as well yeah I mean the best comparison I can give is uh I’veheard about when calculators first came into uh what came into existenceand then they fought it for a while but then they made papers that were specificallycalculated papers yeah yeah so do you think a similar thing would happen with AI what will be the point like like asin surely if it I mean it depends what you see schoolas well I do think it’s a place where children need to learn Concepts and then process these Concepts and then applythem like if you introduce AI to to Greater next can extend into into itthen they’re not really learning what’s the point of them learning anything in thefirst place well we said earlier right that we still have to be able to process that information rearrange itlike it’s not understand it yeah so that will still be the case like rightnow if you don’t know something you will look on your phone and you look it up yes but then but then apply you’d usethat in whatever you were doing so is it not but but then then as I said like atsome level these AI tools are more intelligent than intelligence is no question like it they they will be moreintelligent than us and they will be more intelligent than the majority of studentsthere’s a question of like why why should I learn this if I don’t need tobecause I like before um you could say oh why should I learnaccountancy I could just pay someone to be my accountant and then you could argue back well that’s going to cost youmoney you know you could have the skills yourself and and save on thatum but now it’s like I don’t need to be an accountant I have an accountant in my hands it’s my free accountant on the AIso I definitely don’t need to learn because that there’s not that barrier to theknowledge now yeah but they’re not just at the next step though because obviously firstfirst everything would have been done by hand then Excel came alongI wouldn’t know people use that for accounting then things like QuickBooks came along which allows people that Iknow to do their accounting without an accountant so if not just the next yeahbut I do think like this because there’s this whole likeI mean if you if you remember the word Luddite so there were luddites and they they fought against like sort of uh massproduction and that kind of thing so now that did actually get rid of a lotof jobs and stuff either now a very human thing is intelligence right that’sfundamentally what we do so we are essentially replacing ourselves that’s the one thing thatwe’re good at and that’s the one thing that separates us out so I think this is fundamentally it’s not just a tool itgoes beyond just being at all like at some level like I said it wouldn’tAI will I don’t think we’ll ever be sent in I don’t think that’s our biggest issue Ithink it’s the fact that it will like even if you’re a topMIT graduate or something like that like you will not have much purpose becauseyou’re all your brains and stuff like that can be you can be outwitted by a bya machine and ultimately that will leave a lot of people jobless like so Ido think I mean I don’t know if you’ve seen that um especially as we’re all doing likeoffice work like a lot like the world has become more office centered right sothere’s less like agricultural you know people working working in their in the fields and stuff like that yeah andthose have become mechanized as well um I think this is very very different fromany other technological technological umuh development it it’s it’s not like the internet the internet created more jobs and morefields of work and stuff like that this is actually we’re replacing humans and what humanscan do and what makes humans unique so I I think this is fundamentally differentfrom anything that we’ve had before how would it affect education thoughum things like essays for example Howards wherever if we’re wise we’ll just keep it asI I think we should keep it as we should keep some of the traditional aspects toeducation and also I think our view on education needs to be a bit different becauseI think we view a lot of people view education just as it means to an end ina very utilitarian way that’s never how I’ve seen education I always kind ofthink there’s a sake of learning something for this able to learn something so it’s because that’s what we do we learnstuff why would you fight that that’s human nature and then in developing an AI tool thatcan take away that need to learn stuff we’re ultimately extracting our valueso I think there is if we’re wise we’ll probably fight against it in some way especially withour education system and we discourage it yeah but convenience is alwaysthe winner like convenience always wins out but it doesn’t mean it should nobody will yeahit always does he always go always go back to the analogy of you could have expensive over here over ear headphonesthey’re all wide and stuff or you’d reach for your airpods like yeah peoplealways go for convenience every single time the convenience solution isn’t necessarily the not necessarily the best but it’s theone that people will go for I think we need to teach kids honestly more aboutthe value of like relying on themselves and why they should care about what they cando in a boxed exam environment exam environment or just survival environment where theyhaven’t got devices or anything or anyone to rely uponum we need to really get kids on board with the idea that that is a useful and valuable thingbecause we can all agree that you should be able to rely on yourself if all thelights go out and you need to know things or you’re stuck in an interviewand you can’t just Google what you want to tell the person interviewing youum if we can start from that position of like getting kids on board like yes we dounderstand you have access to this that and the other but do you agree that you should pursue yourown learning and pursue your own intelligence even though it can all be handed on a plate to youand then move on from there because I think we do have a tendency to likepunish kids or make kids feel bad for technology that’s just been placed intheir lives you know they didn’t okay they might have like begged mum and dad for an iPhone but didn’t create theiPhone like the technology is there there is so yeah we we need to encouragethem to want to learn without technology tofall back on and then teach them from there because I think we have a tendency toyeah force it without that prior conversation it’s like with our calculator a calculator is is just atool because you know what the operations are now have owning a calculator itself isnot enough to answer a question that’s given to you need to know how to use it an AI is fundamentally different wellbecause it can take a question and interpret it for you so the the child hasn’t doesn’t have todo any learning to come up with the correct answer which I think isfundamentally not what we should be doing we should be teaching the principles by which you come to theanswer and also how do you know whether the AI is doing something correctly because youyeah I take it as gospel we know that people do that anywaythey’ve got so how do they develop critical thinking skills or anything like thatif they don’t actually live analog to some extent yeahokay so we’re going to go through the positives and the negatives just to finish off um I’m going to read out each of my

The Good and Bad

bullet points and you can both give me one sentence oneach one maybe some of them are quite self-explanatory so we’ll just see how we do

The positives

so automating repetitive tasks and becoming more and more efficient I mean that is useful yepyeah yeah it’s quite a self-explanatory one how about HealthcareI see it having a big imp impact on diagnosis yepum I know one of our tutors actually he he works for I can’t remember whichcompany again but he he went off to become an he was doing machine learningbefore that so he went off to work for a haircut Healthcare company that works in DiagnosticsI’m you know using these AI tools to actually [Music] find out what people have and predictivemodels so they can work out whether someone will have like cancer or something like that yeahyeah I think AI will really help with like individuals Healthcare and justputting their knowledge into your hands without necessarily relying on a medicalprofessional as much and I do think that you know official diagnosis reallyimportant um we shouldn’t be handing around medication to just anyone without youknow understanding exactly what they need it for but I think having the knowledge in your hands andthe models like you said to track down your symptoms to a diagnosisas opposed to how it used to be which is like Uh something’s wrong with me am Idying answer is yes you probably are um you know if we can make that moresophisticated then people can trust themselves a little bit more in their own research a little bit moreand I don’t think that’s a bad thing our NHS yes especially with the way with thewaiting times that we have we can’t just rely on doctors because we can’t seethem all the time yeah it’s kind of a perfect thing for running millions andmillions of data sets of models to see what outcomes are going to happen the year Diagnosticsdisease progression as well um personalized medication yes becausethey can test like they that but they’ll probably one day be able to like scan you and take your blood and stuff likethat and then just automatically churn our perfect just an MLT yesaccessibility so I’ve seen obviously Apple Vision Pro’s just been released I’m sure they’ll have somethingin there to do with accessibility for people that can’t see or hear very well uh they’ve also they’ve also released auh a voice I think it’s called personalized voice did you see thatno but I mean even I find like when driving or something like that like just using Siri no no this is different sowhat they’ve done is you record 10 to 15 like phrases andif you lose your voice it will speak for you in your voiceokay or if somebody else wants to take your voice yeah uh so that I mean I say oh and itwas pretty impressive from what I saw so how is accessibility going to be affected byAI so like Vision hearing yeah I mean I think technology is all likeadvancements is that in technology is always going to help people that are struggling the most becauseif you’re struggling with accessibility it’s because you can’t access it in thenormal way that everybody else is um that’s that’s kind of like the disabled experience is like you’redisabled because everybody else is able to do this that and the other so iftechnology can creatively create a detour or a alternative planfor somebody who can’t access something in the normal way thenmore and more technology is only going to increase the chances of that yeah that was it bioniconslike I’ve seen some quite a lot of very sophisticated bionic arms and and theyliterally like restore someone’s ability to to to live like a northern like a normalperson so with an arm S I was just thinking of uh I.T crowdwith this space star ordering this helicopter and it’s robots yeah so so yeah I seethat that’s useful but I also I think Elon Musk made that point where if peopleif people own the AI and and it’s open source then that makes a massivedifference to you know our ability to control itum so yeah like having open source solutions for for people is importantbut the more it sort of funnels to the the extremely wealthy ornefarious parties who can use AI for bad things essentiallyum the the more of a problem that’s going to be okay uh we spoke about education but wespoke about it sort of relatively negative so what are the good things that AI could bring just a quick sentence ortwo yeah one of the positive things I was thinking of earlier is umai’s ability to individualize learning and so get right tothe exact level that a student is at and meet them there and give themsteps and stages and rewards andum yeah encouragement that is exactly designed for that individual studentyeah and that’s got to be the best thing we can get out of [Music]um the processing power of AI is yeah it’s gonna Target down to how you you’relearning in that moment as opposed to putting a brush acrossyeah they’re all this age so therefore they learn in this way yeah um yeah it will really kind of get to uhmeet the child where where they’re at um and that could be really good put to good use I think there’s an empoweredteachers a bit and actually reduce their workload significantly butyou know they’re obviously it has to go through their their filtration process so a teacher can recognize when aresource or something isn’t going to be that effective or their students soI think it’ll help them quite a lot in that regard even though he has loads of negative consequences I think if yougive too much AI to children it’s yeah I’d just like to say like one thingas well on the whole technology and education thing so I was working in a school recently and they were likeimplementing more and more technology like data collection on the students progress and you know that’s going to beused to calculate their levels and what they do next and all these sorts of things but I guess the one thing I wouldlike say if I could you know put a halt on technology or AI you know gettinginto education is like to not forget that you know children areyou know they can surprise you like they can make leaps of improvement that youcouldn’t have calculated in a model or they could just be having a really bad day or you know emotions and very humanthings are at play within a school even though we’re building in technology tomake it more efficient and more you know kind of like a a system you know we needto allow room for a child that all of a sudden just walks in one day in acompletely different place in their progress than they were yesterday and we don’t just say or thecomputer said you’re not there so can’t calculate that within my algorithm soyou know that’s the one thing I would want for the education system is tonot lose yeah just have a relative amount of Freedom within the systemsthat they’re creating yeah I think it’s better to think of of children not in terms of sayranking them or that kind of thing it’s better to just think what don’t they know what do they knowand you know focus on what they don’t know and strengthen their rather thanthinking of them as you know this students here and the students here just that that studentknows a lot more than that student so that student needs to learn more but what what do they also know and what dothey also flourish in outside of the box because a student that’s reallystruggling with English and maths might flourish in English Math Sciencein every other language if you consider the fact that they’re really good at playing piano and you encourage thattalent and that boosts their self-esteem and as a result has an impact on all their otherprogress so again like we could work it out through calculations but I doubtthat a technology is going to suggest oh have we asked that child if there isanything that they do enjoy or they are good at yeah and we could help withtheir confidence their confidence might be the issue rather than their ability to process English and mathsproblems at home as well yeah exactly it’s all got to be not left out of theequation okay I’m gonna just go a couple more quick ones I’ve got onpositive so things like security and fraudum because it’s very easy for a model to run through and see anomalies in thatsort of thing well yeah it will I guess it will bring both negatives and positives and they’ll they’ll be righton both sides they’ll be probably massive leaps in Innovation when it comes to security and yeah alsoum for companies that are unable to or struggle to upgradethere to the latest form of like like usum yeah that that could also be an issue both ways yeah yeah and last one socialgood potential to go either way like same with security really yep well I thinkit’ll be a disaster I’ve always thought that oh doomed I remember yeah even when I was uh becauseI was thinking about this problem at school and I was like yeah and like some like because I think I’d heard of The Singularity when I was like 13 14 and Iwas thinking about it I was like yeah this is going to be bad like some someday this willum this will bias in the bot basically if you think about it there’s you knowwe spoke about ants earlier as being this like limited life form that youknow we look above like you know maybe that’s just where we’re heading and like ants still exists yeah they just theyjust do that one thing yeah and maybe we’re just gonna be the answer to AIworld so to a super intelligent like being that came to this earth like an alien or something like thatmaybe maybe that that isn’t the worst thing it might just kind of knock our ego a little bit butprobably it depends if they form us like chickens hopefully they don’t eat us but um theymight just leave us alone like ants I don’t know okay um last thing the negatives so

The Negatives

um this is probably the biggest one where should we start bias of the systems they try and do quick quick fireso yeah bias to the systems preconceived Things based on who’s the person thatcoded them yeah or based on what’s already on the internet whether it’s left or right or whatever it is yeahit’s just a case of who shouts allowed us I think and we already know that it’snot equal in the sense of people have different opinions different things that they care about and maybe not the samevolume of voice in the world yeah so if it’s trolling Google it will bewhatever has the most information on yeah whichever side it is or if it’slike a closed system that somebody’s coded that person who has coded that will have their opinions coded in I’veactually found that chart GPT another thing they’re not that bad in that butthere seems to be a modicum of objectivity in how they’ve you know how they answer questions and that kind ofthing so but yes it can be a problem dependencewe’ve already said the smarter machines make stupider people um so if we depend so heavily on somethingthis with this level of intelligence what would be the consequencesit made me think about like domesticating dogs to like what we did to wolves we made them really stupid anddependent yeah they can’t yeah like if you look at like a chihuahua now theyjust look off at you like please help me kill me like I can’t survive on my ownbut you know they used to be able to before we got in there with our intelligence and people have gotcleverer over time the average IQ has gone up yeah but I feel like we’re withalmost peaked if not already big but then one shape it will go down from this point the only way is down because likewe we needed that intelligence to produce this AI that will replace our intelligence soyeah I suppose but I think people will still do things out of the love of likepeople still learn out of the love of learning even though they don’t need to be a place for there’s still a place forreading books when you can watch a YouTube video like there’s still a place for doing it the old-fashioned way jobdisplacement yeah I think that’ll be a massive problem like I think people really underestimate how bad it will be becauseso many are going to get replaced I remember like I saw a YouTube video andit was just like 20 people but can can chat GPT take their job andI was like well it’s not doing a very good job sorry but I don’t think it’speople who have a tendency to underestimate these these systems and how quickly they can they can changethings with the improvements I will come yeah I do think on an emotional level though they’ll get to a point where youknow there’s a person sitting there CEO of a company that they’ve been buildingover the last couple of years surrounded by the technology that help them to getthere just thinking I’m a bit lonely you know it would have been nice to have done this with a team that I can nowcelebrate with and sail it was hard but we did it you know taking the kind ofI want to say cheetah’s route of like not hiring anyone justimplementing technology wherever you possibly can I don’t know if it will leave people assatisfied as again doing it the old-fashioned way of getting a group ofpeople together to make something happen like efficiency is great but I don’t see usas like fully efficient driven uh things because we would do so manythings differently if we were um yeah I think that would change how wethink we’re forced through like you know theway Society changes and if you want to survive you’ve got to go this way or that way but we’ll ultimately alwayswant to have people around us and community andhuman nature yeah ethical issues so I’m gonna list up afew things privacy weapons like well surveillance thatcomes into privacy uh Healthcare decisions and criminaldecisions like in a court of law thought on any of those wellI think iRobot actually did a really good job of showing some of those that’s all thekids going back to my mind with this stuff it’s like it’s like we learned nothing from watching that filmthe problem is is that regulation moves like a glacier right it’s really slowand I think ultimately the technology will people underestimate what it can dobecause the rate of technological growth tends to be more exponential than linearso um in a relatively short period of time we’ll havetechnology that goes well well beyond what we can regulate andthen you’ll need some AI system to do regulation in order to keep up with it but I don’t think it will beeasy or possible because people actually need to understand the regulations and that kind of thing when they put inplace and the implications of that regulation I think it would be easier to regulate how humans use AI than toregulate AI it’s just going to be a case of like okay go and take everyone’s phones awayfrom them because we can’t we can’t deal with this anymore before we actually find a way to put a fence around thetechnology wow that is until it’s implanted into your into your skull withlike neuralink or whatever it is you can’t you can’t say that away really I think Elon Musk had the bestillustration of the danger of AI and that was it was to do with this seeminglyinnocuous AI that was meant to generate Returns on the stock market yeah yeah soyou just envisioned this this thing and he said okay well what it could do is it could gocould go short the stock market or something like that but and then it could and then go longon uh Tesla sorry well sorry say companieslike um that are specializing in in Warfare so likewas it uh caterpillar yeah is the industrial but you get the idealike I forgot the name of it Industrials and stuff like that it could send some blockage the Lockheed yeah LockheedMartin and those ones um B BAE Systems and those kind of companiesum and then it could send like some kind of threat to Russia or something likethat or to North Korea or something like that and then create a sort of a bit ofa drama between countries and Nations and and then before you know it likeobviously it would produce a massive return on the stock market and thenbecause it will use logic that is not human that’s that’s the thing is like it’s unpredictable what it can do didyou hear about the uh I can’t remember exactly there’s this case of uh as AI weapons and they they uh I can’tremember how it happened but eventually the AI turned the weapons onto the soldiers and killed the five of theminstead because they said well this is easy this is the best route to kill the least amount of people and blah blahblah blah but yeah yeah like it came up with its own yeah you just figured out it used yeah it uses its own like weirdweird computer rational not logic you know what it kind of reminds me of Ikeep thinking of like you know like the genie from Aladdin yeah so it’s like allpowerful can do whatever you want but it’s inch like the person in chargeis the person with the lamp right yeah so whether it wants to or not if it gets in the hands of somebody withevil intentions um then it has to do that if it gets the hand of some hand of somebody that justwants to fall in love with that girl then it’ll do that it’s really like ithas no choice but it’s all powerful at the same time but I think Elon Musk withthat point was actually making the point that even if it weren’t designed seemingly forsomething harmful or if even if you didn’t prompt it to do something harmful it could actually result in a harmful anoops Yeah like as in something that you don’t predict the machine can do yeah but it will because it sees that as thebest solution yeah and all options are on the table like with the human there’s certain things that we just wouldn’t doeven if it was the most efficient efficient way to get what we wantedit might be a different one but it says um an AI drone kills the human operatorduring it’s a simulation yeah but it says the system started realizingthat while they did identify the threat at times the human operator would tell it not to kill the threat but it got itspoints by killing the threat so it killed the operator instead umto calculate things in a certain way like oh you’ll get points if you do thisand minus points if you do that and all of a sudden that becomes the entire world of that machine in that moment andit’s like people just don’t work like that you know you can say to someone oh this is how you earn money and this is how youlose money but they still have human qualities to deter them frombecoming completely yeah but unfortunately I can’t really teach you can’t really teach thatit’s all done on a weighted it gives different thingsdifferent numbers different ways different weights so which is based onthat on the training and everything it learns more and more and this is the only way that it can do it you can’tby the moment anyway can’t possibly have those feelings that that we would have Idon’t I don’t think you ever will it won’t have the feelings that’s the problem that is the problem yeahbig problem we’ll probably leave it there that’s two minutes to go through super intelligence unforeseen consequences well we’ve kindof covered that super intelligence it’s just like it’s so intelligent that you can’t you couldn’t predict what youcould do right so it is stupid yeah it will be too dumb to realize what whatcould what could occur and that’s not like a slight or not intelligence it’sjust like the reality is if you if it kept incrementally improving imaginesomething with like a 4000 IQ what would that even look like let’s just be that’swhat I’m saying once it starts um communicating in a way that we can’t understand anymore then it’s like it’scoming across that point awkward yep well that’s a nice nice Cherry note to

Outro

end the uh and the conversation on um do you want to sign us outso that was about how uh the human race will die out and thanks for watching thanks for watching we’ll see you next time if we’re still here check back in 10 years and yeah I’ll let you know [Music]thank you [Music]