Marketing Buzzword - Machine Learning - Elizabeth Clark

The Marketing Buzzword Podcast on iTunesThe Marketing Buzzword Podcast on Stitcher




Hello again and welcome back to the Marketing Buzzword Podcast!

This is the podcast which helps you to understand what all of these business and marketing buzzwords actually mean, and how they can helpful going forward, and today’s buzzword is “Machine Learning”

I’m your host Ben M Roberts and in this show I am the marketing bee in charge making sure I can get the right guests and ask the right questions to make these words and phrases make sense.

In addition, The Marketing Buzzword Podcast is now powered by Talkative. Talkative is a company that brings live chat, voice calls, video calls and co-browsing together, in one package. This allows you and your business to truly engage with your customers, offer quick and effective resolutions to questions and improve the customer experience. You can find out more at

The Marketing Buzzword Podcast is Powered by

So, how does this show work?

Simply, you, the marketing bees let me know what buzzwords you’ve been hearing, and bring them into the marketing beehive. I then bring on an expert buzzword bee from the field, who helps us identify what these buzzwords actually mean, and whether they are useful or not. Essentially, I want to de-bunk or de-mystify these words and phrases to make the marketing jargon a little easier to understand.

I’m always keen to hear about the buzzwords you are loving or hating right now so please drop me a tweet to @roberts_ben_m or simply use #marketingbuzzword on Twitter & Instagram or if you’d rather you can go on the website and let me know through there, and take a look at the show notes from previous guests.

Then one final thing before we get onto this weeks guest . . . if you are loving the show, please feel free to leave me a review on iTunes! It really does help boost the shows profile and make it easier to keep getting on these expert guests.

Right that’s more than enough about me. It’s time to introduce this weeks guest! This episode’s expert buzzword bee is the the beautiful and intelligent woman,  Elizabeth Clark .

Elizabeth Clark is the CEO and Co-Founder of multi award-winning  Dream Agility, an international conference speaker and bestselling author.

Dream Agility is a Google Premier Partner and one of only 35 Google approved Shopping Partners in the world.  With 2 patents pending, their award winning 2nd generation cloud based Adtech sits between the clients website & Google/Bing/etc. It optimises data ‘on the fly’ and makes net profit related bid recommendations, delivering industry beating uplift for retailers, lead gen, and travel sectors.

Based in Ramsbottom, Paris, Atlanta, Melbourne and Seoul, Dream Agility remains at the forefront of it’s sector – internationally. Often acknowledged for innovation, the company recently won the prestigious 2017 Global Tech Challenge in South Korea and were also named in the MEN’s Top 100 Greater Manchester Tech Companies.

Always looking to develop their software, Dream Agility has the first ever cross-faculty Knowledge Transfer Partnership (research program) between Manchester Business School and University of Manchester, on an AI, machine learning/ neural networks  enhancement piece for their platform.

Elizabeth Clark LinkedIn 

Elizabeth Clark’s twitter

Elizabeth Clark’s website

Enough small talk . . . let’s talk “Machine Learning”

Machine Learning Transcript between Ben M Roberts and Elizabeth Clark

Ben: Hi Elizabeth. And welcome to the podcast.

Elizabeth: Good morning.

Ben: Good morning. I am so excited to get you on and talk about machine learning. It’s such a word and phrase that so used interchangeably with other words and it has so many different connotations and meanings. So I just really was excited, so excited, to get you on to actually bring back some clarity and actually define what it is and what it isn’t. So I’m going to hit you straight up with the first question is, what is it and what does it not. Because I tell you what, I’m confused and if I’m curious, I know I’m not the most incredible person. But I thought I’d like, I’ve got a pretty good grasp of marketing and I still sometimes have no idea what it means so please help me and everyone listening. What is it and what is it not?

Elizabeth: Well, machine learning very, very simply put. It’s just a mixture of computer science, economics and neuroscience. And AI, Artificial intelligence, is just something that does a process. It can do a thing and what’s happening is people in marketing in particular are talking about using AI in Google. Oh yeah, we’re using AI. We’ve got all this AI going on. We can really improve your processes and results with AI. Actually what they’re talking about, It’s just a few scripts. And the issue with running a few scripts is there’s no intelligence there. And what can happen is they can layer up and layer up and layer up. So an example here, there was a client we looked at and they had an agency that was using AI on their account. But when we looked all they were doing was just compounding the scripts. And you start off with something that was, say the original big costs was £3.25 and there would be a big multiplier on that for London 25% on. That we’ll put up it up o £4.06. And that’d be a morning so another 20% on and that would put it up to £4.87. And I’ll tell you what, it’s mobile, let’s put it up another 10%. So all of a sudden, not bid, it’s gone up to £5.38 that’s a 65% increase. And we’d see these, we call them Frankenstein babies, we’d see them everywhere, all through accounts. Google’s got no way to really deal with that. They don’t have an answer for that, but the answer is certainly not layering up scripts and layering up scripts. It’s not AI, it’s the most basic thing that you can do and we need to stop calling it AI because it’s confusing everybody.

Ben: And that’s the thing. It’s confusing to the layman, let alone even people who are working in the industry and it’s one of those things. I just at a huge conference and they were talking about AI and machine learning and bots and all these different terms and how you can automate all these processes. And I’m sat there going  hang on one second. On said machine is this and another said AI is this. But hang on, I could quite easily flip those round and almost. It almost sound like it’s the exact same thing. So how do people know whether they are working with a company that’s actually just layering on scripts or whether it is pure machine learning. How do people differentiate that?

Elizabeth: Well, you can start by looking at what types of people do they employ. We have a very small company and we employ 2 machine learning guys. One of them’s got a PhD in particle physics and he’s worked on the Higgs Boson. So those kind of people are generally associated with machine learning. If you’ve got lots of people who are google qualified, then not associated with machine learning. So the likelihood is they’ve got nothing going on that. We also work with Manchester University, we’ve got the first ever cross faculty KTP. It’s the knowledge transfer partnership, just means it’s a big research project. So we have one professor who deals with machine learning, one professor that deals with neural networks and he’s helping us develop our, Dave projects. We tend to call things funny names when they’re in production and the names stick and then we can’t think of new names for it.  So our visual AI product was called Dave. We’ve got another product called the violator. We got another project called the purger. It sounds dreadful when we’re speaking to our clients, “if we just run the purge on you”.

Ben: That sounds really incredible. Its sounds really worrying but incredible

Elizabeth: Haha. The way Dave works, so one of the big problems at the moments, but particularly with Google. I’m supposed to be speaking next week at SX in San Jose, but unfortunately as you can hear from my chest, I’m too ill to travel. So my husband’s going to go out, my CTO, he’s going out to San Jose and he’s going to speak out there. It was one of the best, sort of awarded talks last time on robots and machine learning. Are the robots coming to take us over? Everyone’s sort of quite terrified about it. The creatives don’t have too much to worry about, but you know, if you’re a sort of bog standard digital marketer, you’ve got an awful lot to worry about.  So, we were looking at feeds at the moment, for example. So the big problem is feeds are difficult to generate, they are inaccurate. And the minute you put that into Google, you start showing for things that are irrelevant and you’re paying for clicks that you don’t want. You get everything, the performance has dropped. So within the industry, what we’re seeing at the moment is, big google shopping campaigns going back in house and moving away for agencies. And agencies who traditionally had very good revenues. Google shopping is kind of king at the moment when it comes to retail and those revenues a diving off a cliff. So really big ones at the moment, Google has a thing amongst the premier partners, they’re working with 10 premier partners who used to have great revenues and who are falling the fastest and they are working with another 10 premier partners who are growing the quickest.  And we’re one of those. So out of those 10 premier partners, the average growth rate. So at the top of the top of growing, on average, about 35% – 39%. And we’re growing at 156%. So it just goes to show you. Google came in to see us and said, “What is your secret sauce?”. The secret sauce is we’re just not doing the stuff that Google tells you to do. We’ve got this layer of machine learning and AI going on. So, basically with Dave he can look at a picture and that he’ll see that it’s a dress, it’s a t shirt, it’s a this or that. But it’s not that straight forward. So, the person in the picture. So it’s going to see the person that’s going to see skin. It’s going to see a handbag. It’s going to see. So it then, starts to work on competencies. In terms of, how confident am I, what I’m looking at. So it brings back all these terms to do with the products. And the idea is we can very quickly build a feed that is very accurate. So we can see it’s address, we can see it’s place. So, we can see it. So another thing is in different countries, they call it different things. So we have clients that work in sort of 20 territories. And in the UK, it will be a playsuit and in Australia, they call it a romper suit. So you also need to be able to change the language and if you’ve got to a younger person searching for something, they’re going to use different terminology to an older person. So you’ve got all these things to consider when you’re trying to build a feed that’s as relevant as possible in order to serve on Google as cheaply as possible.  And Dave does all those sorts of things and one of the things we find interesting with Dave is when he comes back with the results because it’s looking at everything. So it might be a lady with a picture and she’s got a top on and a pair of shorts. But it’s a top we want to concentrate on and it’s bringing everything back. So we do supervised machine learning, there’s two types of machine learning, supervised and unsupervised. And the supervisor stuff is basically where humans checking, what the machine is telling us, to see whether that’s correct or not. And then when he’s being supervised, we were getting too much information coming back from the image, which is confusing for a human being because really we’re only want to talk about the top, not the shores. So we’re gonna want to talk about the top, I don’t want to see terms relating to the top. And so we removed all those out and then a very strange thing happened last week. It was a picture of a model came on and then all of a sudden was a load terms came back, not relating to the product she was wearing. It was anal sex, porn star or porn magazines, the name of a porn producer. We’re looking and saying, “bloody hell. What’s going on here?”. It was a bit of a surprise, but it’s also looking at things the way people have been tagged on the net. So this girl has potentially been starring in something which is a brand reputation issue. So all of a sudden you’ve got, if we can find that information out then so can of the general public. And if you’re a big brand and you’ve got a model who’s very attractive, doing something for you. But she’s also done very lewd in other films, you might not want that in terms of protecting your brand. So it’s developed this, discovered a whole layer of other things going on that we weren’t aware of at the time. And there were some websites associated with it, but nobody wants to click on them. just in case.

Ben: You’ve just dropped so many knowledge  bombs are out there. I’m writing some of this stuff down because I said so many of these things I want to sort of drive in on. I think I want to start  on that last one. And I think it’s fascinating actually, that we all sort of know in this digital world, nothing’s ever private anymore. As soon as you post anything online, even if you’ve got a Facebook account. Let’s be honest, google, Facebook, whoever it is, knows more about you. Probably, you know more about me than I know about myself now at this rate. But it’s like, where does, should people be worried about? Is it, should people be worried about our needs. So I know I’m going slightly off topic, I know bring it back into a marketing perspective in a minute, but is there a worry that people who are actually driving the change with machine learning that actually they are going to invade someone’s privacy? Is there a worry about that and is that something that marketers need to consider in terms of when they’re trying to adopt or build a machine learning platform or work with someone who has a machine learning platform?

Elizabeth: Well, it was a surprise for us. We weren’t expecting that kind of information to come back. But anything you put out on the Internet is all typed up by somebody and at some point it will come back. So,the response to that is don’t put it out there if you don’t want it coming back at some point, don’t put it out there. We had no idea. It certainly wasn’t saying we were trying to look for. If somebody said, here are all our models, try and find the ones that have done porn movies and it’s not something we would have thought about doing. And then we have to also test actually was she in porn movies or was it something to do with the way she was looking or is it associated with somebody who looks a lot like her? So it raised a lot more questions and then we had answers for. Then we said, well, should we find some images of porn stars and put them through the image recognition to say, see what Dave says about it. But we decided we just leave that, we’d park that one there for now. But I think the answer to all these things is,  if you don’t want somebody to know something about something, just don’t stick it on the net.

Ben: And then I think that’s an incredibly good point generally. Whether on this podcast or anywhere else in life. I think that’s just a general great, great rule to put up there. If you don’t want other people, don’t tweet something because someone will say. It will be ultimately saved by a robot or someone somewhere and will always come back and haunt you.

Elizabeth: Years ago they didn’t have huge twitter warehouses, data, houses, of all the tweets. In order to do all the sentiment analysis, you can think, you can try and retrieve things and put you can’t. It’s not easy at all, Once it’s out there. It’s out there. So just don’t do it kids.

Ben: Yeah.  I couldn’t agree more. And then when the, I’ve got so many points, I’ve been writing down from everything you’ve been saying. So when the other things that sort of came to mind is, you talked about sort of the supervised versus unsupervised machine learning. Now, if you implement, sort of a pros and cons approach to either one of those. Now, if you were to have a supervised approach to machine learning, does using machine learning actually save you time? Or is the time almost spent equally? But obviously you get better results than if you were to do this all manually, so do you know what I mean? Because you actually have to spend the time putting all the algorithms together, the process, actually monitoring how the machine, how the bots going trawling through everything and then having to check everything yourself. So you can see that would be quite a time consuming process, but it’s still saving you time on doing it manually.

Elizabeth: There’s absolutely no way we could do this stuff manually. Some of the feeds we’ve got anything from a 1000 products in a feed to 2500000 products in a feed. A human being is not going to be able to go through and look at all those images. But so what happens is you train the bot as it goes along. So he’s looking at that, looking at the images, reviewing the images. We’re saying whether they’ve got it right or not got it right. And it learns from that. So you don’t have to tell us again and again and again. So, we’ve got this big neuron net going on behind what we’re doing. So it’s learning everything we’re telling it so it doesn’t have to be applied again. So obviously that’s going to be a rigorous process because we don’t want a top coming up with porn star at night at or anal sex or something like that.  So, it has to be supervised in that respect. But if you look at the unsupervised systems, they still have a level of supervision over the top. Because there’s a big debate about, you know, are we’re going to build bots and are they going to kill us. And also driver-less cars, you know, the car is taking in so many data points at once, way more than a human being ever could. But then, you have to make the decision if it’s going to crash, does the car choose to crash into the old lady or does the car crash into a lory or something like that. And that’s where the layer of supervision has been placed over the top of it by humans can help make those decisions. So that the machines making most of the decisions most of the time. But when it comes to a critical things, that layer of supervision and that helps them inform those decisions.

Ben: It’s funny, we literally had almost a really identical conversation in the office this morning. And we were talking about how with driverless cars so basically we’re an office with a load of devs that are building our products. And they talking about actually how, with ethics and stuff as automated driverless cars come come to pass. If a developer, for example, puts a couple of lines of code in and that car ends up killing someone. Can have a developer then be done from manslaughter because they’ve entered a line of code in it. It is a fascinating world and I didn’t have the answer for that, but something I’m just got my brain ticking overdrive. I need to find out the answer to it.

Elizabeth: Well, I mean it wouldn’t be so much. The developers put a couple of lines of coding because the machine is learning all the time. You put an input in and there’s different layers in the middle and then as an output. So a developer could put something in there that’s a bit dodgy, but the machine would realize it’s just a stupid developer, right? It’s not going to be hanging on one line of code

Ben: Well, that was a simplistic example, but the principle of it was where we’re going in the future. But then moving, so a little bit further from that then is, with machine learning, is it something that anyone can do? Is it accessible? Or are their really quite big barriers to entry? So if a small medium sized business would want to try and develop some of their own machine learning or programs, how technically savvy do you have to pick? So I’m assuming, it’s not something that you couldn’t just pick up and run with.

Elizabeth: Well I’ve got my colleague, my VP of technology, sitting across from me going, “Oh yes”,  and nodding away at me. But then Daniel’s very clever. So we like say, we’ve gone to the lengths of actually getting involved with the university because we want the best people around on this. So we want the support of the university, but we’ve actually, both Daniel and my CTO Glenn,  they went and got themselves as massive book. A massive book with tiny writing in and they started reading it. And when the two guys being university came in, we produced this book. “We’re reading this, we’re up to page 700 and whatever”. And they said, “Oh yeah,that’s what we actually teach on that machine learning course, but we only use a couple of chapters of it”. So already we’re streets ahead of what they’re teaching university.  But there in lies another problem. So, if you’d gone to a machine learning conference a few years back and we’ve just been full of academics. But now when you go, it’s full of people from the big four. It’s full of people from Facebook and Google. We’re going to have a problem because the likes of Google and Facebook are buying up an entire departments of machine learning people. They’re just taking them all, hiring a lot of them and if they’ve hired them, who’s going to teach the next generation? So, you know, there’s a disparity going on here. It’s not very fair, but the guys who got money could just sweep it and by all these people up. We’re very lucky at Manchester. We’ve got one of the biggest machine learning and departments in the country in terms of the experts and people who’ve got round here. But then, you know, Turing started up here and went to the computer. So we’ve got this huge heritage and long may it last. But yeah, in theory anybody could do it. But certainly when I was watching Glenn read the book, I had no desire.

Ben: I think that’s an incredible point in all areas. Not even if you look at things like construction, engineering, business. There’s so many things were academia hasn’t quite kept up with the real world. And this is when we’re almost like academia was a head of a practical application. Now practical application and capitalism is almost overtaken the academia, which is almost a bit bizarre in modern world. But it’s like, how does. I lost my train there.

Elizabeth: Happens all the time. Just have a moment.

Ben: We’re back. We’re 100% in the game now. Right, so is there a risk then if machine learning almost petering out? Or they’re being a real stunt in the development of machine learning because of these big tech companies buying out the almost the teachers of the next generation? Are we going to hit a point then potentially in your opinion, where we suddenly go because the rate of acceleration is huge at the moment. The growth and the development speed as with technology is incredible. Are we gonna are we close to a plateau or do you think we are still a fair way away from a plateau in terms of where we’re going with machine learning?

Elizabeth: There’s not enough companies at the moment want too. I think people would like to employ these people in house, but they don’t really know what to do with them or how you do it. Because certainly when we started down this route we said, this is such a massive problem. We cannot get the best quality feeds we need and for human being to do this is just ridiculous. But the only way we’re going to have to do it, it’s actually by looking at the image. And it was a big decision to make because we got very good links with the university anyway. And we had a chat to them about it and said, “is there something you could help with?”. And they said, “well, yeah, it’s definitely something that can help with”. But there’s two, you know, it’s half neural networks and half machine learning. And we’re, like I said, we’re very lucky that we’ve got those resources in Manchester, say they’re happy to support us. But we found actually that on doing the project, it’s the more, some of the commercial side of things and more the economic side of things that are actually driving the priorities. So they’re looking at solutions. How do we solve this problem? Blah blah blah.  And we’re saying actually, but the commercial application of this, we just want it to be mainly right, not a 100% right. It’s just to be mainly right because we don’t live in a perfect world. You know, when it comes to advertising online is have you got the stock, in stock. Is it at the right price? So we developed this other bit of a huge bit of kit that’s coming out imminently. It’s basically a big dashboard. So when people ring us up and say what’s happening to my sales have dropped off a cliff, what’s going on? And it’s always the 80:20 rule. So I mean people themselves can do some very, very basic stuff. So if you parade through everything, then you have a good feel for what’s going on. So you look at the best selling products annual, so look at the products that drive at the most cost and you get a good feel for what’s going on in business. Because nobody can look at that many skews at once. We’re looking at that. It’s never our technology because to start with when you build something, you think, it’s the platform it’s the platform. I don’t even realise it’s not a platform at all. It’s always with the retailer. And it’s either because they’ve run out of stock and they don’t know they’ve run out of stock because the digital marketing guys, they’re getting slammed all the time.  You’ve got to advertise, you’ve got to advertise. We do come across some digital marketing people who we’ve shown our technology to. And they have said, ” I like to work in real time”. I said, “well, you might like to work in real time, but Google isn’t working real time”. You’ll have an attribute window. And your attributes window could be a fast fashion. It could be sold to three days, something like that or longer. And like, “no, no, no, no, I know when things are selling”. I think, “you don’t, you’re clearly an idiot”. They’ve got no idea when things are selling because they’d no idea how long in terms of attribution, what’s actually working, what’s not. So what they’re doing is just guessing that sitting there and pressing buttons and guessing. But what we’re looking at is actually has the product been in stock because it might be in stock now, but it may have been out of stock. So the dashboards show that that product’s only been in stock 45% of the time, so the other the 55% of the time, the best selling products have been available or not been available. And that can have a massive impact on sales. And then the other thing that impacts it is pricing. So how’s pricing, have you been competitively priced and has it now completely dropped off a cliff. In which case we can send them an alert to say we’re going to drop the bid on this. Because you’re not competitively priced anywhere where all you’re doing is harrowing money and is haemorrhaging your sales. And that’s not a good situation for you. So we’re using the machine basically to advise human beings of actions that they can take. Actions that they would never have been, it just wouldn’t been possible for them to go and look at all that amount of data to establish that. And that’s where we really need to use machines to our advantage, we’ve got a skill shortage in this area at the moment. That’s going to flip on his head very shortly, once everything gets out there and everyone’s sort of using machine learning as standard. And they’re using platforms as standard. They’re using tech as standard and the application of humans reduces massively

Ben: So in terms of the future development then. Where else is machine learning currently being applied or wherever the big areas of growth in terms of machine learning being used in general businesses. We focused quite a lot onto advertising and e-commerce for. What are the other applications? So I know there are huge amount. I’d be interested to hear your thoughts on where else machine learning can be integrated into businesses.

Elizabeth: Fraud is a good place to look. What’s going on in how to spot patterns in banking and things like that. Where you do a thing where you’ve done three micro-payments and all of a sudden your credit card stopped. They’re looking at the performance of your account, where you normally do things, how you normally do things. Is there anything looking strange in your account? If you just had a rule that said if you have three tiny little micro payments go through, we’re going to close the account down, then that would be lots of unhappy people. Particularly if you make lots of micro payments. So it’s using that kind of data and personalising it to you to make informed decisions.

Ben: I think that’s a really incredible point actually. Machine learning, I think it can be quite easily construed as it all being big data, big volumes. Actually, machine learning can be applied to almost to the individuals to make sure that businesses can tailor exactly what they’re offering for the exact individual. So the right product for the right person at the right time, the right price point, in the right place. And it’s that sort of application which people seem to almost neglect and, or don’t understand. They can offer that power.

Elizabeth: Yeah. And you’re exactly right there. So you hit the nail on the head in terms of product recommendations. You go on a website, you pick a pair of trousers. And it says here are five more trousers like the pair you’ve picked. I just picked a pair, I don’t want another pair of trousers. Well, how about I shirt to go with it. I don’t like this type shirt, oh they don’t like those types of shirts. Maybe they’ll like these types of shirts. Is that sort of intelligent learning about you and your behavior while your on there. I think that also, you know, if you look at things like search boxes. Google pulled out search boxes last year, on site search. And nothing’s really stepped up to replace it. So when we’re looking at feeds, we’re looking end to end. So that data is current. People, when they go onto websites, they’re expected to be able to search the way the web designer wants to make them to search. And that’s a horrible experience.  That’s why Google is so popular. There’s nothing other than a search box, no faceted nav that was hated that went years ago. So, you know, we need to be looking at giving people the best experience possible. And in your search box, if somebody searched for certain thing, you want to make sure that: A)It’s in stock; B)It’s a good price; C)It’s the right size and we’re not doing that at the moment. But it’s completely possible because we have the technology to do it. When we run a experiments on clients, such as search boxes, we’ve been able to discover, you know, hundreds of thousands of pounds worth of uplift. Just by using the feed rather than using such technology on its own. So there’s all sorts of applicationsout there, for people to make more money, have better experiences getting the right thing. If you look at the likes of Tesco and Debenhams. They’ve loaded themselves up as marketplaces and it’s really hard to find anything in there anymore.

Ben: I get it.  Again, it’s almost like it almost helps negate that paradox of choice. Where you’ve got so many products out there, so many companies, so many offerings, too many choices, even just down to I want to dress. And it’s at, well I’m not saying that I want to a dress. But in the context, I did want to a dress, there’s nothing wrong with that. If I wanted to dress and what sort of style. What do I want. Straps, no straps, blue, black, gold or don’t even get me on. How actually to ask you about that, how your machine learning copes with things like blue and black dresses in white and gold anyway? But yet that reducing that paradox of choice, where people are going, what do I want? I just want to dress, but what colors, sizes, shapes. Whether it sort of tucks everything in. Whether it has holes in it. All these amazing sort of fashion trends. Actually it’s really difficult. I mean eight sauce is a huge example. You go in there, look for my girlfriend to get her something, bloody hell. I stared at it for five minutes when I can’t, I can’t do this anymore.

Elizabeth: It’s overwhelming, isn’t it? In terms of the point you made on colour before. When people searched for colour, they searched for very basic terms and the issue then have is the colour of your products. How do you get those two relate to those very basic terms? So we built a massive colour library using all the tones and everything. So actually the machine can look at it and with a degree of competence, say which colour we’re going to match this into. But again, you know, you might have an older audience who might call green aqua or pink Fuchsia. Because that was a fashionable term when they were younger and they continue to stick with those terms. In which case you can relabel them and do anything you want. So it’s all, it’s all personalised and it’s a, you know, machine learning is great. I love it. It will remove all that sort of overwhelmingness from people. I think if you just want to see the blue one. Yeah. Did you realize that paint, you know when they’ve all got these crazy names for cans of paint?

Ben: Yeah. Like rat brown and sunset orange and stuff.

Elizabeth: That apparently is a law. You’re not allowed to call, it’s an IP thing. You’re not allowed to call one paint colour, the same thing as another paint colour. So every paint colour has to have a unique name. It’s just particular to paint. It doesn’t apply to carpets or anything like that. I had no idea where it came from. But when you’re trying to sell paint online and somebody wants yellow paint and you’ve called it moon dust, whatever. It’s very difficult because you’ve got to call it something different to every other paint that’s ever been produced. Who invented the craziness, blah. Again, you could just plug that through Dave and he’ll tell you what exactly is. So that when it shows, it will show in for the, for the correct thing. Because obviously when Google classifies things, they have these hidden fields, for the colour and material and things like that. So go in there and actually show for the right thing. So you’re gonna in breach of calling your paint the same color as somebody else’s pain.

Ben: It  was amazing and it is one thing that people have taken away to forget all the machine learning and forget all that. Just know that you can’t call the same paint, two different things. So the final thing I’m going to quiz you on, though. It’s almost like a bit of fun just to finish this talk about machine learning is, how does the machine differentiate? We knew there was a big thing between whether that dress was black and blue or white and gold, everything. How would you want machine be able or do with machines be able to do that? Because that’s almost like a human perspective on what it was. How long would the machine cope with things like that that are all sort of almost perspective and people will use it completely weirdly. Did you join some? I’m trying to.

Speaker 2: I’m just saying that’s the human problem. That’s not machine problem. The machine. We’re looking at it and use confidence intervals to decide what color it is or isn’t, so they will suffer from perspectives

Ben: what it was blue and black anyway, but that’s. That’s the end of the dressing, but it never stops. It never stops. Anyway, and before I let you go because you’ve provided some incredible insights. Real some real clarity on what machine learning is, but I can’t let you go yet because I need to find out what are the buzzwords, marketing buzzwords or business buzz words and phrases you either loving or hating right now. What sort of really getting on your gears are? What is one that you are absolutely loving at the moment?

Elizabeth: When I’m hating it at the moment is snowflake. I was speaking at a conference yesterday or international women’s Day and the term came up and we’ve actually discussed it. Anime thing before I go home, because I knew this term was coming up and I wasn’t sure whether, um, it was something that meant the same thing to all people as young workforce here. I’m, I’m the mother of two of them. Um, that 23 and 21 of the ages of my two kids and working the business. 10 only five, so he’s not coming out, although he did ask for a car for Christmas so he come to work with his big brother and sister. But I sort of said to them, what do you make the term snowflakes, you something you find offensive, but one of them, so just looked quickly at me, mom, I think that’s the term for the white weedy guys in jail.

Elizabeth: I said, how would you know that she’s up and watching prison break. Oh, excellent. So, um, are informed, you know, media thing, but it is a, it’s a term that’s being bandied around a lot at the moment. So snowflake feminism, um, Jennifer, what’s the facebook wearing that dress out in the cold the other day and all the feminist started on us saying she was doing a really bad thing for women by going outside in a beautiful dress and she said, you know, I choose to get outside in the beautiful dress. Um, then that’s me, I, you know, stopped being sexist and then it sort of evolved into a snowflake sexism and it’s difficult to know where to go with it. But to me, I think it’s, it defines a generation of people who’ve been brought up with a sense of entitlement that going to university where traditionally you went to university to, you know, spread your wings, it’s going with your parents get drunk every night, but up dats and now it’s a place you go where people don’t say anything offensive, they have safe zones, everything’s OK at these poor people are just not going to be prepared for the workplace at all because it’s just not like that here.

Elizabeth: Um, so that’s the one I would like it to settle down and to actually mean something in particular. So if I’m calling someone up snowflake, I know that I’m really insulting them or I know that it just means something.

Ben: Yeah, it’s weird, isn’t it? And it gives me is when there’s. Tim’s almost like out of the podcast is try and bring back some meanings, but snowflake is one that almost has almost so many connotations and I’ve seen it being used in that sort of prison, that prison scenario before. But I’ve also used it to her terms. No, no, no, no, no, no, don’t, don’t, don’t, don’t make me have a bad reputation on here. No one is snowflake generation in terms of, um, in terms of people just in like a right to be insulted by of things where they feel like they can just, uh, they’ve been to university, they just insulted by everything that everything’s a drama. But I also look at from the other plan to go, actually this is a generation where they funded for the first time people are going, no, it’s just not OK to be racist to someone and that that would make me a snowflake because I don’t find a black jokes.

Ben: Funny or is it because then I think it’s, it’s a fascinating subject and one that I maybe maybe I can address on the, on on the podcast soon if it becomes more used, more and more in the marketing or business world are really good ones come up pretty well. Thank you so much for coming on today, Elizabeth. It’s been an absolute pleasure, so you’re given is just be absolutely incredible. I mean I’ve made incredible number of notes myself and I think it’s, it’s a fascinating topic and one that I think is only set to get bigger and bigger and even more meaningful impact upon the marketing and business space. Thank you very much for. Thank you.



Please enter your comment!
Please enter your name here