How we can stay ahead of AI | Martin Warner | TEDxSwansea

[Music] [Music] so how can we feel optimistic in the world of artificial general intelligence that is the question for today so let me set a challenge for us all and we won't go too far back but to the end of uh 2022 open ey ey broke ground or broke Silence with GPT and it was the first time that we could look at kind of machine learning and understand that machines could understand human context and this had profound consequences so much that it got the industry rallying tons of investment the large players started to break ground

as well with their own chat Bots but there was something profound about it and I started to think about it for a long time here's someone in the field for 20 years and was doing deep learning and at the same time thought oh my God now What's the progress going to look like as a result of this now knee-jerk reaction to the first player breaking ground what would it mean what progress could we expect how fast would it happen would there be a replacement effect I heard that that earlier could it replace Humanity possibly and

then I I thought well actually what what's an interesting way to look at that is relevance you know what what can we say about humans that have held us you you in front or given us an advantage over the years and we're going explore the fact that history is repeating itself even though it's about to speed up a lot lot more um the great replacement so having thought about this it would be remiss of me not to just mention a few Concepts when we see lots of lots of technological change and when we apply scale

to it we start to think of the large ramifications and that instead of the computer side or the algorithmic side of our brain we normally go to protection and feelings so we become defensive about this idea of well we're can to lose our jobs um you know ultimately who is going to be the producer of meaning who's going to govern us um you know is this the end of humanity well I humbly think that that's that's not the case in fact I think there's an incredible opportunity to look at the replacement F and say that

actually history is going to show us that it hasn't actually replaced jobs that nothing's replaced humanity and actually we're very much in a pole position but s what is the what is the challenge for us and that's that it's how we partner with AI systems how we all recognize that we've probably got a a a bigger role than we think today in working with this sped up information craze that we're going to go through and that partnership and placing ourself at the right point is the simple message behind this presentation and we'll get to that

next so some reasons for optimism uh I think as I look at a lot of the conversations around AI I think there are now we're starting to get out of that kind of worried concerned defensive phase to more optimistic presentations but we get a lot of this I wouldn't call them conspiracy I I consider them polar kind of conversations around the dreams of what could happen but if we were to go back and I'll take you back to the '90s I was actually part of this movement I'm turning 53 and I was big into automation

where software was you know automating us out of a job it was taking data it was speeding up roles and people said oh my good this we're going to lose our job here what what are we going to do but a macro level it was clear that if you look certainly in the western world where there is a similar economic system that we could recognize there was no effect on unemployment whatsoever in fact humans went and repurposed themselves and had wonderful careers after the automation phase in fact automation didn't affect every industry but it did

have a significant effect and is a worthily replacement example let me take you to the early 2000s so you know we get we got the worldwide web we got search and then we got e-commerce and a way to think about is we we got access to this vast pool of information everything was going online and then Google and others came along and they organized it for us so that we could search it and then we started to transact on that information products and services through e-commerce there was another replacement effect here everything's going online what

will we do about education what will we do about brick and mortar structures um you know our Shops going to disappear on Main Street or High Street we're you know depending we're here High highry I guess and what are we going to do about this idea of of of living in this new world that's online or or in the cloud but actually what we found was that the High Street found its purpose still and it's been much more of a slow gradual process indeed U certain things Vinyl Records books you know we've found other ways

to to distribute them and acquire them but equally we found other reasons to have physical premises so that gradual process did not create a replacement effect so if we sum that up if Humanity 1.0 was this kind of age of information abundance if you think about the web we never I'm old enough to remember the A to Z and yellow pages and all these other things that were out there we could not fathom that we could have all that information and still be relevant but if if that's where we're coming from We're certainly at the

age of information meaning and that's because now machines can understand human context and this is profound because in in in an extended conversation we would see now Robots coming onto the classrooms into Care Homes you know on the streets if we look at workflow automation that's been around for for 60 years and now we add AI agents which I'm going to come on to which is essentially a program that's going to run a bunch of tasks for us and and answer questions that we raise we're going to get to a point where the life is

going to speed up and enrich our lives and we have to question where we want to be so quick sum up of what we're up against and I'm going to Sprint through this and I think you've got some context of it today artificial gender intelligence is a rather loose uh set of definitions espoused by different parts of the world so for for the sake of clarity I'll give a very quick definition if we if we consider this as a as a benchmark a place to start the artificial general intelligence is when machines can equivocally or

outpace human thought they can reason and think like humans so with that definition I'd like to Leap Forward and just quickly say well where will we start to see the very first form of general intelligence we might say that's the pause for opportunity and rightfully concern and this is where a lot of the conversations actually start and it's because it's an it's an accepted view I think that what we see today is the very very very very beginning of public Ai and its limitations chatbot limitations and you're noticing every week we're reading another announcement another

large language model another blah blah blah that's creating something in the world of AI so general intelligence allows us to espouse a a wider Theory so people use it as a beginning point or a stepping off point uh to the next part of the world so domain-based agis is essentially a specialty field where whether it's an industry a sub-industry or an interested group it is essentially a highly focused database or dynamic database let's call it a large language model that will that will undertake very specific task to conquer particular questions within its field of specialty

this is profound because today we have these very very wide large language models GPT would be an example and there are a number of of others out there with Google and perplexity and some of these others that that are out there anthropic Microsoft Etc this is important because now what we can do is we can Target communities within a field of specialism that are contribu into this large language model and we can start to compact that that Centric nature of what we call providing guard rails or business rules so we can think much harder and

reason longer at coming up with some of these Solutions now tag on a program think of it as a curator indeed it's just a script A UI and a set of workflows that we call an AI agent and they are tasked with answering our questions and automating or carrying out tasks and doing that without our knowledge they may do that in the background and here is perhaps the most profound opportunity that we've ever seen so in the field of medicine there there'll be vertical lmms and IR agents and then there'll be the same in automotive

sector the same in financial services and the list goes on indeed vertical llms and AI agents are now spawning in the tens of thousands soon hundreds of thousands and indeed quite possibly millions in these verticals Industries sub Industries or interested groups or parties so we're going to have a profound opportunity to get to this kind of you know generalized intelligence so the importance of this is a few things one is we're going through the fastest technological change we've ever ever seen and by the way this is not a technical presentation this is a presentation that

affects everyone including my mother how do you now look at this this world and say well if it keeps speeding up where where does it end we're going to get to a point of ubiquity where speed is no longer longer a meaningful measurement it's not important it's actually far more about placement it's where the human endeavor takes us and how we want to partner with AI let me use um an example quickly if you can indulge me in the world of human productivity an area that I'm I'm focused a lot on recently if you take

your daily smart tools we all have smartphones I'm I'm sure and we think of a calendar um your notes or journaling tool SMS email tasks projects these tools are kind of static tools until we give it an input it doesn't really process anything but now imagine that an AI system can take that community that vast information pool and analyze your life and come up with productivity Solutions and make you increasingly digitally aware the profound consequences of that is that AI can think fast it can see around corners it can understand what the collective what we

think and design solutions for you so an example would be your alarm is brought forward half an hour in the morning you wake up and you feel it's a little early and you realize that AI has reset your alarm woke you up because it understands on Tuesday Wednesday Thursday you go to work and on Monday and Friday you work at home it's your day to go to work and it realizes there's traffic on Route so it wakes you up early it tells you there's traffic on Route and it arranges your new transportation and prepares you

for work we're just human this is this is one limitation that we can't plan 24 hours just like business doesn't sleep AI doesn't sleep while the system is on so this idea that we can build vertical llms with an agent on top there to serve our our our interest and and and drive us forward is going to create instant productivity opportunities we're about as a world to become not just more digitally aware but incredibly productive as a result of this contextual change I make the claim I think more people are coming around to it but

I I really think it's here I see it from from the world I'm in and I see it from my kind of natural learnings as I'm a curious curious person the AGI is actually here at a domain based level domain-based AGI is here and actually it can already outpace human thought except for the comedy it's not it's not a doom and gloom conversation and actually I don't think it's the end of the human race far far from it and it is about optimism it can outpace human thought so what is the point and there's a

key message in this slide and for this presentation and it's a simple one and it's where we place ourself in the process so if we think about this model there is all these communities spawning large and small vertical LMS all all over Humanity tackling you know work and and life uh challenges and planetary challenges like let's say the climate while all this is happening all these agents are coming up with ideas and solutions as that information is under as an AI agent undertakes the task and gives you an answer it's our job to receive that

information and work with the AI agent and be placed in front of it to set the next requirements for that AI agent and that continual partnership has some very similar um characteristics to other transformational change that we've seen before back in the automation era back in the cloud era of the early 2000s it is going to be about us being able to position ourselves so that we are looking in at this information this vast sped up world that we live in that's also going to become again vastly automated and say well okay this is wonderful

we've got all this information how do we now pass that next set of requirements and that's what's going to be important I can tell you life will be fast much faster than it is today and I'm adamant that um it will feel more automated but I think it can be extremely fulfilling and we are as relevant as always and so if you're a doctor I'll hit some of you here for sure but a practitioner an entrepreneur you know a researcher an educator a manufacturer it doesn't matter a student we're all relevant because we have skills

and we have experiences and that's what humanity is made up of and actually if you think about what AI is looking at it's looking at the sum Collective of human knowledge and what's interesting about that is that we're whever we are there's going to be potentially millions of vertical llms and AI agents and all of our skills are likely to find many many of these pools of of information and these agents where you can place yourself using that information and evolving that information and being part of that progress and I happen to think that that's

an absolute certainty for those that want want to do that there'll be small and large systems as I mentioned for everyone don't worry that's in abundance the one thing we know is there'll be enough computational power there'll be enough databases or what we might call large language models and AI agents are going to start writing themselves and we're not going to worry about the the task maker or the person that answers our question and there's an important Point here and it's a real simple one that uh AI systems aren't sentient they can't feel like us

they can't deal with gray judgment and a single AI system is actually going to spend less life less time on this planet than most humans so we have a huge role to play in the aftermath of AI so how can we be part of this journey and this is perhaps one of the most important points because I know that my my my mother probably would not um care too much but actually we all have to jump on the train and not a stopping train because it's a competitive world and it's speeding up we all want

to get on that train we want to lean forward lean in and actually analyze the information and put yourself in front of that AI agent in your part of the world and help guide the next set of require requirements some of the greatest achievements curing cancer extending life expectancy food water prosperity it's all there to be had in this next wave That's What I Call Humanity 2.0 thank you [Applause] human