foreign [Music] [Music] feature Young Jamaicans who are shooting for the Stars I'm your host Margaret Boyne today we explore the fascinating world of Robotics and artificial intelligence AI once considered futuristic is now one of the most impactful and game-changing Technologies of our time my guest the CEO of stone Technologies limited a Robotics and artificial intelligence company my guest is Matthew Stone so what about jobs that will stayed in what about the trades job then you see here in the back in the day that boy learned a trade oh I think those persons will have that edge no well it's turning out like getting those jobs out to me is has proven to be a lot harder than we thought it would be yeah because you know getting our machine to because the human body is a very um fascinating thing like dexterity and how the brain controls your body um it's not a very easy thing to replicate mine they are the robots are getting better and better so with time even those jobs I think will get you know automated as well so definitely as you know Society goes on like you're going there's going to have to be some sort of upskilling period there's gonna be so it have to be some period where you know some people feel disenfranchised and and you know conversations now need to be had about like how do we kind of sort of upskill these people into you know new jobs or you know educate them master or you can use AI to make other forms of wealth you know not just depending on a job that they've done for like 20 30 years but all right let's see how we can adapt because it's going to require that's the thing with AI it's going to require us as a humans as a species to adapt and improve um that's that's how I kind of view AI way for us as a species to adapt and improve but yeah the idea of like a job existing forever AI is going to hit every industry every industry yes um okay so um AI is sold to us though uh Matthew as something that will make our lives you know better almost like we believe in a like a Utopia but then we've been hearing a lot know about these dangers of um of yeah I talk a little about that about the concerns all right so there are some valid concerns and there are some not so valid concerned I feel like you know because of the whole hype with AI a lot of people are taking advantage of it so that you know to get attention and news get into the newspapers yeah kill us all destroys you need to stop AI you know it's the same people making AI telling you that we need to stop it um I think in the near term there are some possible ill effects that you need to look out for so for example bias with our AI systems uh depending on how these systems are trained uh it can be biased towards certain things um you know it could be by race it could be um by you know gender gender yeah exactly stuff like that um or you know inter even in terms of like the content it recommends to people you know we've seen where these social media sites have actually created more polarized um Society um so that that's that's definitely something that we we need to definitely look out for because you know Bad actors can use these AI Technologies now and like because of the content generating um capabilities now can easily send out like misinformation like no other time before and easily like influence societies you know win an election with that if if necessary um I think also yeah automation is something that we need to look at that's one possible you know ill effect um when you consider how we as a society are going to adapt to you know AI taking jobs yes over the years you know new jobs have been created but that's not a certainty you know we don't know how quickly the society is going to be able to adapt you know some Saturday it is me adapt quickly some may not um so what's going to happen in that case um we also have to we're going into a future where it's going to be hard for us to tell what's real or fake because that's generating images videos literally look just like real life and sound like the person and all that exactly and even scammers are using AI technology right now um fake um phone calls and I'm sick I need some money and it sounds just like a real person and people get fleeced out of money so they're there like dangerous like ill effects that are happening right now um that we need to be careful of I think that the more kind of sci-fi futuristic um concerns are the whole is the whole idea of you know the robots uh turning on us and taking on overheads no but what what what uh Matthew okay I laugh about it you know because I'm not gonna I mean I used to the Creator on this thing here coming out sounding scared I I mean I don't know that's not possible or it's never going to happen or you know humans are gonna find somewhere you know obviously everything is in a realm of possibility and I think that there's a possibility of that happening I don't think it's in a near term and that's something we really need to worry about per se but you know it it is reasonable to think like in the near future you know as is machines get more capable they get more intelligent how much power do we give these machines and you know what can they do with this poll um you know what things do we give it control do we give it control of our utilities our financial systems if they get sort of this ability to be self-aware and you know conscious like you know what could they possibly do to us the species um the truth is nobody really knows exactly what's gonna happen yeah everybody can sort of I guess like oh this is gonna happen all right but it's it's really people just kind of guessing some people just kind of going off of what they see in movies I I don't know much I don't know listen to me that Google that Google um I don't know I can't have a thing twice you know yeah I think a lot of the news media kind of um sensationalizes because I heard what you're saying um they're kind of sensationalize what they said to kind of make it sound more like but um yeah definitely there are dangerous possible dangers to AI even some that are happening right now in the near term like the bias ethics you know automation that we do need to be concerned about um and you know it it you know it does make the question of you know in the near future what what what are the possibilities of these AI systems you know is there a possibility that they could possibly you know turn on us and end up in some sort of AI apocalyptic scenario um it's a possibility um but definitely it's it's one of those things where um when it happens you might not even see when it comes like it no and then everything is happening so fast to Matthew I don't know yeah because the speed at which it the speed at which argument that other people make is you know if that does happen right if we somehow create technology that can think for itself and maybe have the ability to turn against those we will as a species will figure out ways to mitigate that risk and we'll create a awkward there but there is a possibility because of how quickly these these Technologies like Advance exponentially we might not have time to even do anything about like once again let me say then we we might not even see when it's happening it might even do it in stealth and you might not even know when it's happening so and and they are making some projections like I mean like 20 30 35 24 Singularity I don't know I I don't know much with me mostly overreact but I don't know right the singularity which um well people there's scientists call records while I had predicted that that the year 2030 would have essentially had a machine that can equal the intelligence of any human on the planet and then that would create like an intelligent explosion where in a smarter version of that machine smarter version smarter version to get some sort of super intelligent machine that some like God like machine that you can't even comprehend its intelligence um so yeah that that period before get that that so-called AGI are artificial general intelligence um there's some researchers that say that's more just like again like a Sci-Fi thing but essentially that is what all of these researchers that's what they're working on that's that's why they're working towards they are as much as they say what we're trying to solve problems and all of that stuff right they're really trying to like see if they can create that you know that that intelligent um think on the level of the human or greater maybe even have some sort of free Consciousness like they are actively looking or working towards it and you know it makes the question you know should we be trying to do that um what's the possible consequences of that um you know are we putting up you know guardrails to guard against the possibilities um in my opinion it's it's moving so fast like trying to regulate it no it is probably not gonna work because tomorrow Genie is over the bottle the Genies like because even if you set up a lot in your country sir we're gonna stop China is going to continue working on it exactly so there's a race is that race you know or go ahead exactly you have to continue running it is the wrong people having control of these things oh you can just say all right let's just stop guys like somebody else is going to continue racing you don't want the wrong people being ahead of the park and that's the problem but with the release a chat GPT though um Matthew to me it kind of seemed like them just throw it out on on us right now and then try and figure it out after oh this thing really work or you know it really boosts to the profile of open AI um slid them to being one of yeah you know probably that next big tech company but at the same time um it's well it's actually improving the system and we have more people using it they can get more feedback right yeah it's actually like one big experiment we're just the Lab Rats um kind of you you're using that all that data to improve the system so that's the that's the thinking behind um you know releasing chart GPT Google has been a bit more cautious because I mean they have a bottom line to protect given that they're a big um conglomerate or you know they've been a bit more cautious about releasing their version of CHA GPT Bard Bard yeah but what I've seen um you know definitely from what I've seen in the communities this idea like just one company is gonna have the the only chat butter this company is just gonna have the best chat but forever like yeah yeah um even in the open source Community like people are creating um things of similar um so yeah everybody's gonna get access to this everybody can be their own version of it yes not to mention some techie like you win in them garage uh it it's opened up uh all you know host of possibilities um Society [Music] um in terms of chat GPT um Matthew um how was it trained or what was it trained on right so child GPT um so you can split that up into your chat you know GPT so chat from chat bots so you guys are probably use chatbots for a long time the GPT is also generative pre-trained Transformer for the T for the Transformer is a type of machine learning model machine learning modeling is something that takes in data makes predictions now the Transformer is a type of machine learning model that predicts sequences of text so for example I said something like the boy took up blank bottle and I asked you okay what should you put in that blank his or her you probably say his now you might be wondering okay how did you know that you have to put his in that blank well if you look further back into the sentence you know that boy was said so that determined the gender of the article is it's so it's it's sort of how GPT learns how to generate text it uses a pattern of the text coming before it and learn statistical correlations of in a particular context what text should come next so it's really just a very very powerful um autocomplete essentially how it was trained literally just give it a bunch of data on internet so text from Wikipedia text from social media text anyway anywhere literally feed it to GPT the books as well they have books humans so every like they go through PDFs anything that has ticks defeated to GPT what GPT essentially learns is based like it sees patterns of how text is generally um created and it literally learns that same same thing so seeing so much text same so much sentences it learns um patterns and correlations and figure okay these string this string of texts usually follows this so most likely I'm gonna predict this word and then you can like okay these set of um text most likely I need to predict this word and it just keeps doing that um yeah and that's where the pre-trading comes in so it first tries to learn how to generate text and from that you can use it on different tasks like question answering things like summarization summarizing a paragraph stuff like that yeah that's that's the only stream so so which means though then that it's it's also trained on you know like that's a questionable sources then like another key part in its training version is something but really because this has existed for a while you know like training machine learning models with that like giving it a lot of things and trying to learn a sequence of text now the thing that really set it apart is they use a technique called reinforcement learning with human feedback so essentially to find the model what they did was let the model like generate some text and they would have humans actually rate it so let's say the human would say the text that the model generated and say okay this sounds good and I learned from that feedback uh to basically generate better zoning text the problem with that and it's actually one of the reasons why it hallucinates you probably heard it before so it makes stuff up because yes yeah there's no source of true humans are creating the text they're just kind of going off okay what sounds good sounds good or it sounds plausible doesn't mean that it's actually a factual or right so that's actually one of the reasons why um GPT actually if you've if you've ever experienced before but literally you will make stuff up it will even speak reference and it sounds so plausible and they really carefully like wait yeah because I I I I asked it to write something on my husband you know and trust me so my information was very inaccurate it's not even going to Kingston college so I'm like really no so if I didn't know that I would think that this is true you know oh boy but that has that has implication though um Matthew 4 oh do we know what is true then what about truth that that I mean that's a that's a question that's even before GPT people yeah so so imagine no exactly okay I have everybody even saying like you have my truth and your truth yeah it's why you have to be very careful when using these tools especially kids um because it will make stuff up and it will sound plausible but you just have to be very careful about what we do you know check against sources but yeah that you know what is true like um that's the big question I hope you know that someone is is really telling you the um what what they're telling you is [ __ ] um yeah that's a big thing that we're going to have to battle with as a society going forward like what is true what is fake what is real what is our life you know it's gonna be hard to tell the difference all right this is this is a hard question though you think um yeah I will become sentient I think that's possible wait what do you think based on what you've been reading I hope it I hope it does it doesn't dry I hope it doesn't bye I want I I because of my Curious mind I I would love to to see that um but you know obviously implications of that yeah at the same time are great but yeah I I think so I think um I think it it even if it it's not exactly it will definitely very closely approximate yeah um why anyway I don't want to cause too much Panic though on Matthew so um so let's go back to you what has been the most valuable lesson you've learned that you want to pass on as a entrepreneur I think the biggest thing that really the most common theme that has come up is just like don't give up like they're gonna be times where like everybody is kind of it feels like the whole world is against you um nobody believes in your idea um you know things just aren't going right like you try everything and you get rejected here you feel like listen like the worst thing you can do is give up like just just keep going on um you have to understand that you know nothing lasts forever so what you know all this pain that you're going through right now it's not gonna last forever some point you're gonna have some form of success you just you just take it by take it day by day you know find something that you really love to do um you know find a process that works for you and just take it day by day by day but yeah the thing you can do is give up like never give up that's my biggest advice oh okay very good um can you leave um any advice to any young persons who might be interested in pursuing Robotics are artificial intelligence yeah my advice would be um just have generally a curious mind um it may seem like very complicated on the face of it but you know even with the same example where literally just Googled how to make a game and I just tried it like just try like literally um you know it could be literally as easy as a Google search you just attempted I think that's really the key to be you know successful in this field like you have to have a curious mind you have to be someone that you're not afraid of taking on new challenges uh you know you just try and that it you're going to feel you're going to be bad at you're going to suck at it when you just start off but you know the longer you you do something the more time you put into something you're going to get better to start to master it you have to kind of understand that first of all that yes you're going to fail yes you're going to um not be so good at it at first but with time you're going to get better um yeah that's the biggest advice I give just yes do it let's try it like it's possible just try it you never know like trust me you never know what spots all right so what do you what do you do for fun um that's a good question some robotics robotics business well actually all right so you know on the weekends I play I play football um yeah when I when I can or when I have time uh you know I watch I play I watch sports as well um you know I like to watch like documentaries and I like reading um I mean to be honest like most a lot of my time has been like you know working on projects which I am fine for yeah outside of that yeah I'd say that like uh I play football um personally I don't think AI is a problem though but just Humanity that's what we should be worried about really yeah okay um it was great talking to you but I I enjoyed this this this conversation and I can't see that Jamaica will be needing your skills and your expertise you know very soon yeah so I hope you are ready for you know I hope you're ready for it yeah this is us that's what I'm working towards yeah man and I'll be looking at looking out for the the research and Innovation Innovative things that you'll be coming up with yeah and it was nice having you [Music] thank you [Music]
2023-07-01