Working with AI (Artificial Intelligence)

Internet DNA Podcast

How does technology - AI (Artificial Intelligence) and ML (Machine Learning) - effect the way we work not just today but in the future (sorry farmers you get a bit of a bad wrap here). What if colloquialism effected driverless cars, the main frame computer keeps raising his (it's) head as picking the perfect pattern may actually just be a lottery and guerrilla marketing, anxious lemurs, meaning, purpose... all in a day’s podcast.

Transcription

(this transcription is written by robots… so don’t be surprised!)

I would meet Abby this week. We're going to be talking about lucky with AI as you pointed out as well, machine learning and artificial intelligence

and how, how are we going to make cross shells redundant? No, you see that as not the case and that's scaremongering. It's a bit like falling over the cliff with Brexit. I'm sure there's some scaremongering fag in that all the jobs that we lose, we're gonna have to gain new jobs to, let's say that we're going to lose 15 to 20% of the workforce because machines can improve it, replace it. We've got to employ about that sort of amount as well. If you read some of the research, it's very similar. The loss and the gain. Now, admittedly they're different skills and so you might be losing certain type of person and gaining a different type of person. But the clever businesses will be training their people to start to be the ones that will be looking after the machines because if they've done it to the human, then they're going to know the pain points, how the machines can make it better and they're going to be the best people to a help develop them and be managed to them. So we're not all going to be after job.

Well I think we need to look at the industrial revolution when we're looking at this, which is a vast amount of jobs did go like farmers and people that push things up and down. Um, people that lifted milling stuff up. And if you think about how much manpower was needed before we had engines, you basically needed a few horses in the loads and loads of men to do anything. And then you create an engine and suddenly you don't need that many people or horses to do those things anymore. To the kind of point where we are now, where the tractors and combines are so unbelievably expensive, but you only need two guys to basically harvest the whole of America.

Well also like we've created a whole new billion dollar industry in exercising because we're not having to do that physicality in our jobs anymore.

Yeah. And also now people are building tractors, so it's not like the jobs disappear, they just move. I'm not necessarily sure that, did you call them the clever companies? I think basically new industries spring up to service either the additional leisure that we now have because we machines are doing so much of the work for us or creating those machines,

additional companies, they are the ones that with the right innovation already have the materials and the information. So they're the ones that can transform and be better than all these startups that go, oh, we've got an idea, we can do it. But they haven't got any of the history and data that the older companies had. So it's not always, oh, let's just start a new company.

No, not always, but I don't think many farms train people how to build tractors. I mean in a funny way they did because the people working on the farms had to learn how to maintain the tractors and so there is crossover there. Yeah. Generally these jobs go to new industries. Historically. It's not like the old farmer bill suddenly goes right and we're going to teach you out make tracks. I don't think that's how it works. Track to companies then compete and it's all about innovation and probably as a farmer you don't have the head space to innovate on tractors. You're trying to get your tonnage per acre up.

Well that is another place for AI to help out. Then

it is helping our farming AI and ml is actually quite a thing, isn't it? Most tractors on big farms now the bloke in the cab is pretty much not doing anything.

He never was. He was listening to his music and yeah,

but you know in the old days when we were young, we used to have a little Massey Ferguson. There wasn't anything automated about that. Right. It just went up and down a field and you had to steer it and know how deep to have the plow and all that was actually managed by human. Now it's all managed by computers and lasers and GPS, so the person in the cab is really like a pilot in an aircraft is just making sure the systems are working as expected rather than actually controlling anything. I'm probably overgeneralizing, I'm not a farmer myself and I think in places like America and Australia where the space is so much bigger, you start to allow for massive amounts of automation where you probably only need one driver and then a bunch of slave harvesters and tractors and they can do the whole job because they basically drive in a straight line from Mexico to Canada.

Let's go back to what is the definition of AI, artificial intelligence and what is the definition of machine learning? Are they the same

that very similar in a sense, machine learning is basically pattern recognition. So you present a computer with a vast amount of data and you ask it to use that data to learn so you're not telling it in a thing or you might give it some basic check actives. There was one game and they just said all you have to do is win and then they just let it go. It's probably something like Dota or something like

and then humans realize that they can even learn from the machines because there was not a way that a human would think, but it was very clever.

I think recently it beat the best humans by had a slight advantage which is it could see the whole map at once. Whereas humans have to scroll through naked. You ever see a bit of the macro at a time and that came from machine learning and I think a lot of that is pattern recognition and understanding how to build up rules. AI is more about then taking those patterns and applying them in a more what we would call traditional thinking sort of service. There's

two types of AI, isn't it? Those, the narrow AI that you have a car brain and all that car brain can do is run the car. If you use that same car brain and told it to go and run something financial, it would be as thick as anything. AI At the moment is all narrow. It's all made for a specific job so it can't take over the world. And then you have AI itself, which is this suppose into brain that can be like a human and can learn to think like a and can therefore have a thought feeling might be an interesting way to use it but apply itself to anything. So at the moment we only really have narrow AI and we're using those in many.

Yeah. So broad AI includes machine learning because obviously they both do bought car ECU system or even these self driving cars. They work to some very clear rules as to what they can do and they all set up for those rules.

I bet. I think the self driving tools are constantly learning from the information they're getting in the way that they're driving and more importantly, the way other people are driving

individually or in a main frame. As in do they send that data back to something because obviously the problem with

it must be a mainframe. Otherwise it would be like colloquial language. Well that's wrong. Room would be good.

Yeah, exactly. And like a car that I'd learnt in Rome, which be absolutely dangerous as hell in England [inaudible]

she had one that I learned in the states. You're just get to around back and have a panic attack.

Yeah. I think true AI would be that each car has its own, probably is connected into the network, so the data is shared, but it makes it sane decision, I don't think right there. That's a very real issue

with machine learning, isn't it? That the machines will learn bias and bias is the one thing that we are desperately trying to get away from, but how do you,

why are we trying to go away from it? Yeah,

because the bias is yes of your man. You'll probably employ men, not women, or if you're white, then you're more likely to employ white people. Yeah, so no unconscious bias within humans that we are therefore by the data passing that bias onto computers.

Right. If a computer was to look at outcome and to say, well, actually when I look at employing people based on that performance, that bias is not a bad bias for them to have. Surely

No, as long as it's the right bias.

So what you're saying is we need to free it from their creators bias, but not free them from bias in itself.

Yes, exactly. We need to frame it from the unconscious bias, I think is what we're talking

because otherwise you've just got a lottery machine. If nothing's allowed, any bias whatsoever, it's then just random lottery.

This is worrying me about the week before last when we were talking about decision making and how it's God, the mainframe computer would decide who I was marrying for the greater good. Now, if I'm putting out as this luxury and he's going, Oh, well today, let's spin the ball.

Well, I've found from my machine learning is it actually doesn't matter as long as I tell you that this is the right person, this is the right person and that's about as best or outcome as you can expect. And I think you're probably going to get into places where that is true, where the computer goes. Actually there is no real pattern other than belief and therefore as long as they believe this is the right answer, of course I was going to go.

But if you're just saying belief, how would the computer to portray that which he come back to you and go, well it doesn't matter. 50 50 or would she go choose a a? I think he's happy because he's sure, but what make him choose a, I say him see unconscious bias. What would make the computer choose a, when it knew that it really didn't make any difference? How does it,

I think it would [inaudible] to use a random number generator, wouldn't it? As in if it realized that it really doesn't matter, it might as well just generate any

to you and say it doesn't matter. Or would it still give you the answer because you just want to not answer.

If it told you it doesn't matter, you would lack the belief that that relationship would work. If it told you, I have selected number 64 this is the best of all possible outcomes for you, so you better make it work. It's more likely to work than if it went, I've just randomly poked a chat. He may or may not read it fully.

Isn't that extraordinary though? It's completely the way you're told. If I randomly picked a chap cause I've no idea it's going to fail. Whereas if you're told this is the one person of you, you will work through any issues for the greater good. If you're ready to go to the top because you've been told that this is right.

Yeah. When I went back into Shaman and medicine men, don't worry.

Convince that we're filling people with the confidence of working with AI right now.

First of all, you're making an assumption that the computer realizes that there is no choice as in it is just random. I think in certain things that is probably true. I mean when you look at relationships prior to the idea of love and choosing your own marriage, marriage has worked and lasted. I'm not suggesting for a second they weren't in many cases unpleasant violence, but I think that's still probably the case now and I don't have the figures to hand of whether that's more or less true.

My orange, I have a marriage for love.

Exactly. Obviously from a western point of view we go, oh, there's no choice in that, but I don't know if he measured it by outcome, whether actually it's any better to choose it yourself or to have it arranged for you. It feels odd for me to say that because obviously I've been brought up believing that self determination is a very good thing, but the more you talk about it before you go, is it really though, am I just making lots of really bad choices and actually somebody else could make those choices for me and I could just be a slave to the system and live a happy life with no meaning without any idea that I'm actually in control.

You see with their meaning. That's interesting thing because the thing that was a lot of, well, midlife crisis is about what's my purpose, what's my meaning and what is it really? I mean if you take that to the complete conclusion, it's

that your life is utterly observed and meaningless. That's true though, isn't it?

So your meeting is making sure that your children are okay and that they grow up to carry on. But I wonder whether computers taking more decisions would make us feel like, whoa, life is even less meaningful so therefore would make us less happy and most suicides and things because you're powerless. And I think humans need a purpose even if it is,

oh, they need a purpose

that for ourselves. And that's why we created God created the fact that we need a purpose for ourselves or whether

we need a purpose. I don't know if we need meaning. If it was explained to you very early on that your life is utterly meaningless outside the fact that you're going to experience for a period of time life, which, and that's

between a purpose and meaning.

Well, your purpose may be to make a better world for those around you. That's a purpose. Meaning is in the longer term or overall when we look at the universe, what difference did you make or what was the point of your life? Well, probably none much like an aunt, you know, if you said Trinidad, uh, what was the meaning of your life in the greater context of the universe and that,

yeah, we've got a, I'm busy.

Yeah, I'm busy. I'm making a better world for the ants around me. And that's right.

Well that's for the greater good. They've already got the machines, didn't they? Yeah. Okay. Talking about events you just threw into last week, something about upolu event. Yes.

Well it's the blue and it's William Gibson, the blue Ansary. So that's pattern recognition, spook country and zero our, I believe it's called. So they run by Schubert's, his big end and he runs a company called blue. And so I was talking about it in that sort of odd marketing way, Rosario. And so I was knowing that. I thought you had said that spook country was one of your favorite books. I thought you would pick up. Yeah, but it's called the type of company was, it was a marketing company. He himself was interested in effectively what used to be called guerrilla marketing, like different forms of viral dissemination of branded ideas. How about that? That's why he was in patent recognition. He's so interested in the film series because it's creating this massive following and it is for nothing and he's interested in that

and it's a big country. It was the locomotive art 3d uh, well the virtual reality out there, you had to put on your glasses and then you could see the art, but no one else could.

Yeah. And how you could use that to change people's experiences of places and things.

Yes. Which I think is brilliant and will happen soon.

It's happening in many ways without the VR glasses, but people are starting to brand spaces for you. So I walked through the broad gates center every day. I'm pretty much every month there'd be another set of stuff that, and usually you had to use an app to sort of interact with it, but it just made for an interesting walk to work because you'd go, what's this funny thing here? And then you'd read about it and then you'd look at it in the app. So it just changed your journey to work. And obviously that's being branded by broad gate and so they're looking for a kind of general feeling that the broad gate, the center is a cool place to work in an interesting way. And it's quite [inaudible] that kind of branding of spaces that I think is happening, especially in the stations. Actually.

You've got quite a big captive audience, haven't you?

I know, but I went to kings cross or some pancreas. I didn't even recognize it when I used to go to King's cross in the 90s I mean it was a hell hole. It dark and dirty. You didn't know where anything was. You couldn't find a ticket. That's, you know, they really have changed. It's quite amazing. Just still have to pay for the they, they've stopped it in Liverpool Street or in pain now. It just stopped all toilet to payments. Really. Very good thing. I think what was happening was everyone was blocking the lose on the train, probably flushing the loo at the station.

Just want to get back to working with AI for a moment. So as a business or business owner, AI is touching everything, machine learning, searching everything. And if you haven't started to make changes within your business, then you need to start looking for them. Now what is the process of finding out how you can slowly and in a small way start to integrate some forms of machine learning AI so that you don't get completely left behind when a startup goes with in your industry and takes off

when you're fairly small, really you need to buy in those services. Whether those are marketing automation services, whether those are auto profiling and segmentation, bending on what your business is, but your probably getting on with your business. And so what you don't want to do is now have to learn a huge amount about machine learning. How to teach machines. You probably don't want to learn a load about AI, but you can buy stuff off the shelf either from existing companies or you can go to cloud services and they allow you to do all sorts of a manner of a thing.

Yeah, I agree. There are plenty of different types of AI that you can just buy and source into your business, but you'd need to recognize where it can help you

depending on your business. But if we're gonna imagine that most businesses to some form of marketing, then really what's interesting is to track your users through your different channels. So through your social and your web and all the different places that they go. And hopefully if you can do any physical outlets that you have, then you can get machine learning to say, well, people that look like this and do these types of things tend to be the sorts of people that will buy more than once from you. And these sort of people are the sort of people who only buy once and these sort of people, they pretty much don't buy from you unless there's something very, very specific and then you can change your marketing to those different types of people. Cause you know kind of what their trigger points aren't. I mean marketing is all about efficiency. It's about sending the least amount of stuff to the least amount of people for the most amount of return. If you could send out five letters to five people and make five sales, that's way better than sending 500 letters to 500 people and getting five sales.

And that's the sort of area that platforms are absolutely. Therefore in sales is the same. That pipeline and that understanding that so and so that may have contacted you here has also been looking at you on social media or they'd be looking at your competitors on social media. And it's the data, it's bringing all the data together and the AI is able to make sense of it for you.

And then the other side is customer service. So if you have customer service, you can start to use machine learning to actually understand what are the things about our product that lead to customer service issues and how do we now design around them to reduce, because obviously that's a cost. You've got people sitting there. So it's about getting machines to look at patterns in any type of interaction people have with you. Yeah, exactly. So where do we lose time? Where do we gain time? It's all about, you know, if we were going to be really honest about any of this stuff, it's about maximizing efficiency or making optimal situations throughout the process from everything from inception to production to sales to after sales.

I think the other place to look as well, where most industries will find that they might get left behind if they don't go, that is making their products smart objects are quite cheap to produce now because of technology and therefore why making that object smart, however big or small it is suddenly makes it more valuable. So whether that smart is the fact that it's sending information back to you to make the product better, or whether it's smart in that it's giving information to the end user. It could be anything. It could be a baby tie that's telling you the sleep pattern of your baby. It could.

Well, we'll, we'll, we'll, we'll, we'll, we'll, how's that helping anybody?

I don't know. It can make you just worry a bit more about the sleep pattern.

Yeah, exactly. Maybe you could worry a bit more about another thing. This is the danger, isn't it? Which is that we create all these things that basically just turn us into utterly anxious Lemurs, frightened delivery thing. Well, Lima's fairly nervous creatures. I think they are. I think they get eating a lot. I think it turns us all into these lemurs. Yes. I think motherhood is an absolute crash course in anxiety and anything that can stop mothers from being as anxious about their children who are probably designed to survive. Most of the things that they're going to go through would help. Maybe less data. Maybe just a, your kid's fine, don't you worry. Carry on. Dave

sending back information on the quality of the water so that you could lower the chemicals or make sure that the water is in a good state.

Okay. But I don't want it sending it to me.

No, you don't want to know. That's where I'm saying that the product can be smart because it can send back to the water company or the product.

Cause I once had a fridge that used to beep when it was too hot and that used to annoy me because if, yeah, exactly. And so I don't need to be involved in this. I mean I always remember a star trek episode where they were in some battle and the guy on the helm goes, captain, the Helm's going unresponsive and the captain goes, can you compensate? And he goes, yes. And you just think, well that conversation just didn't need to happen. You are at the helm. You can compensate it going unresponsive. I've got other things to think about.

Yeah. But that was a film. So use the watcher needed to know. Otherwise you might not even have caught on that bit. You might. Yeah.

Otherwise, you might have not thought, oh, they're in terrible peril. Get you shot out in the middle of space. And I understood why from a film narrative for it, they did that. But in reality, if I was the captain of that ship, that helmsmen would be put on janitorial duty for the rest of his career because that's not what the caps infill his name with. Well, you're in a life and death situation may be filling my head with unneeded information is probably not the way forward.

This is why we're talking about smart technologies. It doesn't need to tell you it just does it.

But a guy I'll send to battle drones and then that sort of the beginning of the robots taking over, isn't it?

Well No, I wasn't [inaudible]. I can name you an awful lot of really, really used lists and not very smart products. In fact, there's a whole catalog called innovation something or other where you have cats with a box on their feet and dogs. I'm sure it does. Anyway. So by making these things smart, yes, you're going to get a lot of uselessly smart, smartly stupid. But as an industry and working with AI, I feel that that's going to be the boom, that it's not just a product, it's a smart product and all those data that therefore is going to be generated where the machine learning and the AI is also making

decisions for you. We're back to the beginning. Okay.

Making sense of the vast amount of information that is coming back from all these people that have bought your products. Big Help. Help. Help Mrs Megan's, his fridge is too hot.

Yes. Let's turn it down then.

That's, that's done it. They're great. Great. Okay. I'll turn it down.

Make it beep at four in the morning cause that's really, really annoying. But in the end you're going to get into this loop where we were at the beginning, which is everything's already decided. Everything for you, especially since I've expressed a preference not to be told about things that you can deal with already so very soon. What decisions do I actually make and I think to put a positive spin on, oh I think it is. I think people need to get away from the idea that them being in control helps. Anything that actually just letting go is probably a good life lesson.

No way back to illusions which back again and then little limpet letting go and how much pressure it was when he did.

Yeah, and I think that's true that your life is absurd. Stop trying to control it and just live in the moment as close as possible as you can given the constraints of the modern world. Hopefully that's what AI and ml will allow us to do.

I've got no more to say. I would love to segue that back into something sensible, but ah, if I am a business owner starting out in business, how can I therefore make myself still interesting for the jobs market? What should I be doing to make sure that I don't lose my job? Is he actually, this is what I think is really interesting. AI will replace jobs, which means that the companies will be making the same amount of money or more, and I should only need to spend half a day per the day that I was spending before. Therefore, I should be able to work for the same amount of money from nine til 12 and then the afternoon, I can look after my children while my husband goes to work, swap over, or I can

go and do some volunteer work, or I can do my art. Or if I really want loads of money, I can go and get a whole nother job if I wish to give anybody and not just business owners, anybody, some advice you need to learn cloud technologies because that's where it's all happening. It's where all the ml is happening. It's where all the AI is happening. It's got the scale to do it next week. We should talk about cloud technology. Yeah, a box of Lego Saturday.

We're going to have to leave it there, but I think technology is next week. Enjoy the ride.

—

Dan & Abi work, talk & dream in tech. If you would like to discuss any speaking opportunity contact us.