Artificial Intelligence and The Marketing Industry: The Threat, the Opportunity, and the Reality.

Editor's Notes

• #3: I haven’t attempted to study Artificial Intelligence. Instead, like many of you here, I shape marketing strategies for some of Scotland’s biggest brands…
• #4: A snapshot of whom are just here. One of whom I’m particularly happy to say we’re standing in the office of now. Wire are a creative communications agency – covering everything from PR and marketing to design, brand development, digital marketing and experiential marketing. As you can imagine, with an offering that wide, we’ve a fair few skillsets in our midst. But there’s one thing that underpins every single one….
• #5: It might therefore seem an odd choice of topic to talk about for us – we’re not a media buying agency making the transition to programmatic – nor are we planning on trading our designers for bots. But, in our eyes, that’s what makes it even more important for us to talk about – the AI revolution is just as important to our skills as any other industry – there’s threats, opportunities, and a current reality. And, in some regards, it’s already here. Whether we like it or not. But…
• #6: If you’re not sure why we should care. Take a look at search interest over the last 5 years – there’s spikes the whole way through, but interest over the last year has been rapidly growing at rates not seen recently. 300% increase from Feb 2013 until now. Regardless of the reason for searching, however, many of the results you’ll find today can look just a little bit like this…
• #7: One side = terror and fear The other = the belief that AI is no more than an extended practical joke. For every article stressing the threat the technology faces us with, there’s an argument that AI will never happen, is overhyped and, even sometimes, a joke. But this isn’t the first time we’ve seen reactions like this…
• #8: Anybody remember this? Believe it or not, the world wide web is just 28 years old. And, a lot like the issue we’re facing at the moment, was originally seen by many as both a huge risk and hugely overestimated at the same time. While, at not too dissimilar a time, others were fearing the end of the world as we know it – from the threat of governments utilizing the internet to destroy civilisations (sound familiar) or the millennium bug out to destroy everything as we know it.
• #9: "no online database will replace your daily newspaper, no CD-ROM can take the place of a competent teacher and no computer network will change the way government works.“ The internet is clearly one of the most important parts of our lives today. So could AI really go the same? Today I’ll chat you through both sides of the argument and lead you through what I think are the key takeaways for us, as marketing professionals, actually are – away from the headlines and using actionable fact. But before we begin – I want to ask, quite possibly the most important question I’ll ask today…
• #10: What, actually, is AI? Hands up if you think you’ve got an exact definition. Well, I’m definitely stealing you after as having read countless books, thousands of online articles, and listened to enough podcasts to drown out the sound of an old firm match – I still haven’t round a specialist who can agree. In use since 1956, just saying the words out loud can bring conflict to the calmest. However most standard definitions go along the lines of…
• #11: The most accepted definition tends to run along these lines
• #12: Hopefully not all humans. But this raises a bigger question…
• #13: Simply saying a machine has artificial intelligence if it does something that would make a human look smart doesn’t take into consideration what humans actually are. If I were to stand here and work out an equation like this right from the top of my head – you’d likely say I was pretty smart. But, as much as I’d love to show my old maths teacher that his work wasn’t all in vein…I can’t. But I can get my phone out and work it out on a shiny calculator which let’s face it, still isn’t that impressive – for me or the calculator. Quite simply…
• #14: Intelligence is a confusing idea – and one that even the smartest of humans ironically can’t agree upon. How on earth can we even begin to decide what it means for machines when we don’t understand what it means for humans. IQs and degrees = useless. Can have someone who struggles with numbers, but can make art. Someone who struggles with letters but can strategise like nothing else –intelligence in humans isn’t quantifiable or easily measured. So, where does that leave us?
• #15: Has anyone ever heard of Go? Over 2,500 years old, it’s known as the ‘4th art’ in China and is considered the ultimate game of wit, strategy and analytical thinking – traits that, naturally, we associate with humankind. However Google, being Google, didn’t want that to stay the case… Google says they built the program to show how computers can be trained to think like humans. Strictly speaking, that’s not really true. The game-playing robot is an example of what, in AI many would call…
• #17: Narrow AI – that is a computer program designed to focus on one, narrow specific task.
• #18: Has anybody ever used any of these tools? chat bots, facial recognition, speech recognition or programs like IBM’s Watson? For those who say yes, you’ve used AI. However, for those who say no…
• #19: Hands up if you’ve ever used this tool? Facebook’s newsfeed algorithm uses something called machine learning – a form of narrow AI - that actively adapts from the data it receives to make a possible prediction or ‘educated’ assumption on how it feels users will interact with it. So, next time you’re seeing a lot of cat videos…don’t blame the digital managers.   Narrow AI has allowed marketers to explore new territories in their campaigns. One of the most popular and accessible methods being chat bots – a first step for many brands to dip their toes in. But how does it work in reality? And what are the results?
• #20: Meet Rose. The Cosmopolitan hotel’s personal concierge… Interesting for 2 reasons… Just how sexist it is Shows how Narrow AI is aiding organisations’ relationships with their customers – ironically adding a more personal touch than without it. So, a far cry away from what TV would have us believe AI is. AI that runs more along the lines of this…
• #21: That clip is from the most recent series of Black Mirror and is perhaps the most recognizable form of AI to most of us – an intelligent being with pro-human skills and – most importantly – it’s own agenda. Seemingly to rid the earth of humans. It’s the type of AI that dominates the headlines and the type of AI we’re most used to seeing in movies –and what it represents is something we tend to call general AI..
• #22: Which summed up in a sentence means the ability to complete any task – not just one – at least as well as a human. Far away from the narrow AI we’re used to using (even if we didn’t know it before) it’s not just focused on one task or goal – it can quite literally do anything we can. And most probably better than we can. Argument that this could lead to own goals and consciousness – not doing what we want, but what it wants.
• #23: So, how likely are we to have killer dogs on our doorsteps any time soon? And what are the real threats according to the experts.
• #24: We’re hardly short of experts warning us. Elon Musk believes AI to be the biggest threat currently facing mankind – creating problems in terms of jobs, company power, and quite literally our reasons for being… Power goes to large corporations – like Google, Facebook, Amazon. And the facts support him – 10,000 deep learning experts in world and only 28 companies with more than 10 experts in the field. Campaigned for all AI to be open source – heads up OpenAI.
• #25: While Stephen Hawking believes the birth of a super intelligence will spell the end of our days. Believes a superintelligence will keep improving itself until it’s far superior to us – begging the question what is our purpose on earth, and what use do we serve to his new species? However – as mentioned earlier – general intelligence doesn’t exist as of yet. So what about the threats from the AI we currently know?
• #26: Ben Carter may not be as recognisable as Elon Musk and Stephen Hawking, but he’s a powerhouse in terms of UK marketing with roles behind him including notonthehighstreet, Betfair and his current role as Marketing Director for Just Eat. His main fear lies in what relying on AI means for the future of our industry and the future of young marketers – when intuition and gut feeling are lost in place of statistics alone - rather than intelligent analysis of each. Interestingly the brand has been placing a bigger focus on AI and technology since Carter joined in 2016 – but the expert’s stern words at a recent Campaign briefing suggested that the introduction of automation could put hundreds of jobs at risk – not necessarily in our generation, but the next. And is he right?
• #27: According to a study by Oxford University and Deloitte for the BBC, it’s a bit of a mixed bag as to how likely our jobs are to be automated. The calculation is based on nine key skills required to perform it; social perceptiveness, negotiation, persuasion, assisting and caring for others, originality, fine arts, finger dexterity, manual dexterity and the need to work in a cramped work space. While the 33% figure for creative directors, and account managers may look high – it’s far lower than other industries and still classified as ‘fairly unlikely’ by the study. Again – it all depends on context – likely the industries these professioanls tend to work in -
• #28: So, are we safe? What’s the reality?
• #29: Current AI is currently driven by humans. And, for now at least, that isn’t going to change. While it might have powers far beyond any human capability. It needs humans to find, aggregate and analyse the data – it needs the purpose from us and it needs human programmers to work alongside building its algorithms and understandings. Difference between observation and insight. Therefore our jobs aren’t disappearing any time soon. Instead, what today’s artificial intelligence brings us is something a little more…collaborative – working for humans to help achieve our own goals. Leading some specialists to wonder if it might be better suited to the name…
• #30: Augmented intelligence instead. The Narrow AI we use often faces arguments as to whether it’s truly a form or artificial intelligence at all – and is instead called something else – far separate from the drama we associate with AI. As we understand AI at the moment, it can help us shape data, chat with customers even when no staff are available, and order items from the comfort of your armchair (when your phone’s in the other room). Essentially it’s an extension of ourselves….
• #31: A little like the first industrial revolution. 1857 – 85% workers in argriculture. Now only 4%. Jobs changed dramatically – manpower replaced by machines capable of lifting and doing more than man ever could. Just because they’re stronger than us physically doesn’t mean they turned against us. AI is essentially the second revolution – but rather of physical strength, is focused on mind. Machines that can do more in a mental capacity than we ever could,
• #33: Will stop the clip there. What IBM Watson essentially did was analysed existing trailers to work out the perfect formula for success – and then applied it to Morgan’s trailer. It read moments of fear, moments of happiness, moments of sadness and applied them to the process. IBM Watson did in 24 hours what usually takes a human team 10-30 days. BUT Important to note that was still created with help of HUMAN editor It’s not about losing jobs, so much as appropriating our skills. Gartner estimate that while 1.8 million jobs are expected to be lost, 2.3 million will be created by the AI revolution. It’s not making humankind worthless, but making us more useful. Leading some to believe a correct term could be:
• #34: One of the most time consuming tasks for an editor is filtering through footage for editing – particularly if they haven’t been involved in the pre-production. Tools like this free up that time – cutting a process allows us to step back and concentrate on the things we’re good at. What could this mean for the wider workforce? The average worker spends 28% of the working week managing email and nearly 20% looking for information internally – what could you do with 50% of the day back. Quite simply, one way we could refer to AI is as a congnitive multiplier. It allows us to step back, and concentrate on the things we’re really good at. Humankind is good at a lot of things, but one I think I think we excel in at any other is being – in a word - inefficient. To invent anything, we need to go through countless trials and make a lot of mistakes. To create art, we need to often experience failure – trying for years to understand our own voice. Human relationships are about conflict and resolution – we fight, we make up, we clash and we click. There’s nothing efficient in anything we consider innately human. AI allows us to embrace that.
• #36: So what about some inspiration about what it can actually offer?
• #37: Even the best data analyst needs to sleep but AI doesn’t – at least not yet. There’s already a host of insights tools that use narrow artificial intelligence – for example programmatic advertising. Word du jour in 2017 – automatic bidding and creative sent to targeted audiences. Facebook’s ad platform - has a learning period which I find incredibly useful. After this time, we can often, with a fair amount of confidence, trust how an advert will perform – allowing us to adjust creative, targeting and landing pages appropriately. These tools are already in play – things we can use today to help shape our campaigns. On that note.
• #38: All of you should be using social listening tools to better understand your audience and what people are saying about your brand. By 2020, its predicted that 80% of consumer journeys will be done without any human interaction. No human element to a campaign – need to create that in our communications. IMPORTANCE OF BRAND. Narrow AI has the capability to track sentiment – and more crucially change in sentiment – much more accurately. You should have social listening tools set up to understand these things – but what happens when your brand isn’t mentioned at all?
• #39: 96% posts on the internet are like these two examples – unbranded. On the one hand you have a positive moment to embrace as a brand – someone (me) at a Marketing Society event connecting with a product. On the other, a negative moment that could be defused early by interaction from a company. AI image recognition can help. Machine learning can process imagery to be able to recognize when people feature it – including logos, patterns and key shapes.
• #40: Used to be only coders who could really use it, but now more openly available to anyone. Last month Google opened up their own Image Recognition platform to anyone – just upload between 20 and 10,000 images to the platform and it will begin to learn the shapes and recognise your product. E.g. meteorologist – recognise clouds and predict weather from them. For brands, we can join in and understand conversations that we’re not necessarily aware of.
• #41: Deep learning can identify trending topics and spot gaps in the market, then suggest new content ideas in response. As addicted as most of us are, we can’t sit on our phone all day long. And sometimes you won’t spot a potential reactive opportunity until it’s too late – to create something special, you need to have time to think and process before a topic has reached peak and is saturated. AI can help – flagging rapidly rising topics, gaps in the market and useful information without the need for a search. However, with tight deadlines you need all the help you can get – and how can you multiply your resource quickly?
• #42: Step in AI again. To launch their Myrai car, Toyota used AI to create a campaign around it. Good because 1) appeals to target audience – innovators interested in tech. 2) The pure scale it could achieve. Saatchi LA wrote 50 scripts based on location, behavioral insights and occupation data – which resulted in over 1,000 videos for Facebook audiences. Imagine asking your content team to make that… The initial idea wouldn’t work for everyone, but the result could – forget one video stretched across countless audiences, and say hello to a more personalised approach.
• #43: Which leads me onto my next point…Personificiation not personalization. We’ve become reliant on data in the last few years – and have created what we know as personalsied marketing. Coke will make a can with my name on, for example. Or an ad might be targeted to a female, interested in technology. But neither of those consider how Melissa – their target – is feeling at that very moment. And who she is away from the demographics. But in 2018, data’s getting a facelift. GDPR will make 80% data obsolete – but AI gives us the opportunity to combat this – analyzing in real-time, understanding how consumers and viewers feel in an individual moment rather than generally. It’s time to look beyond the obvious demographics and start thinking about WHO we’re speaking to, really.
• #44: I’m a female aged 25-34 and I buy makeup, but that doesn’t mean that’s who I am. I have lots of other interests. And in 2018 it’s the brands that realise this that will make the biggest progress. Amazon – 35% total sales from recommended products to buy Netflix – 75% content consumed on channel from recommendations Nice, but not amazing until you consider the algorithm behind it. Not just based on demographics – based on deeper habits – Netflix’s algorithm identifies similar characters, familiar plot lines, actors and directors that a user has shown an affinity to – which is why many of your recommendations may seem far removed from what you’d expect. 1/5 people who watched Stranger Things had never watched anything classified as Horror on Netflix before – but think of its success. Now this might seem well suited to these brands, but not marketers. But we like to think differently and we’re currently developing AI powered ways to bring this to influencer marketing in particular – rather than a cut of those similar interests, we want to find ways to connect with new audiences away from immediate likes and dislikes.
• #45: If you work in an agency, you’ll consider it per-pitch. If you work in-house, you’ll consider it during the recruitment process. Not just a quick look at skills or experience – goes much deeper than that. Needs to be a balance of the right skills, right passions, right processes to make a team. In a small company, you can do it on a fairly emotional level, but what about large organsiations? Or cross-agency teams who have little experience of working together?
• #46: Marcel is an AI program created by Publis Groupe designed to create global teams to work on briefs – from their 80,000 members of staff. Able to create truly global teams that can work together to fuse cultures and new ideas – global work for a global audience. Believe in it so much they’ve pulled out of all marketing including Cannes. People no longer work for companies. They want to work for people – making global connections crucial.
• #48: If I said one of these pieces of art was created by AI, which would you think it was? Hands up for left Hands up for right The answer is both – they’re both entries from the Robot Art awards – where machines are tested in their creative prowess. Now I showed it’s potential in terms of editing clips to make a film but these are something else. The one on the left actually won last year’s competition and $40,000 prize.
• #49: Now as beautiful as they are, not all of us have access or requirement to a machine this powerful to create the art. So… What can you do now to make a difference?
• #50: Now I’m sure most of us have heard of/are terrified by GDPR guidelines coming in this May but if huge fines and platforms banning you aren’t enough to terrify you into sorting through data…the opportunity to maximise on new technologies might. Essentially the narrow AI we currently have access to works on data – even if you’re not thinking of using it yet, getting it in order now could make all the difference when you are. AI = risk of distorting data as it can only understand what it’s fed – for example, it may think women are less suitable to being CEOs because there’s currently less of them. Likewise risk of fake news – by 2022 people in mature economies will consume more false information than true. 90% data made in last year but 80% unstructured – need to structure data. Get it in order.  
• #51: According to Google, 20% of all Google searches were conducted by voice in 2016 – with up to 50% of all queries conducted by voice by 2020. There’s few businesses today that don’t take a firm SEO strategy seriously – but we need to start thinking about how we adapt this for speech recognition. And there’ll be huge differences; from the difference in the way we communicate vocally, to the different onus voice search platforms will place on particular aspects of a brand. Before I explain, I’ll show you a good example of a brand who understand the future of what voice search could mean…
• #52: How many of you are currently investing in Alexa skills? How many of you have a tool that’s up and running now? That needs to change – here’s an example of why. Purina pet video – helps recommend the right dog breed for people looking to buy and gives tips and advice. Not just about immediate searches anymore. SEO strategies currently work towards short and long tail keywords – with the advent of sound, we need to rethink this. This is an example of a brand thinking ahead and understanding the importance of tools like this in the future. As we move forward, it’s not just about Google anymore – Alexa will make purchasing recommendations based on the information it has about you. If it understands you already interact with a brand is instantly going to make it higher.
• #53: It’s ironic that we say the main difference between us and narrow AI lies in that AI doesn’t have an end goal in sight. When sometimes, the best marketers don’t either. Temptation to try new technologies to show you’re innovative – but they only work if they connect if they’re applicable to a bigger aim. When drawing up your marketing plan, think about ‘Low-hanging fruit’. Don’t overcomplicate or be overwhelmed by AI. AI is not going to solve every problem in your organisation. Not yet. It’s still in its infancy. But it is moving quickly – so don’t stand still.   Just ask yourself, what is the intelligence assistance we can offer our customers in their daily tasks? Keep it simple and take baby AI steps in the right direction.
• #54: When the US post office looked to innovate using AI, it wasn’t about finding the reason to do it – it was the right tool for them to use for a much wider purpose. How many times have you gone to post something but not had time to pop to the post office for a stamp? Or ever found the queue within it too long to wait in? It’s AI that genuinely addresses a consumer problem.
• #55: The establishment of AI needs to be supported by company leadership – they must be prepared to fund it and give it sufficient time. Now is the time to share with your boss, you board what business needs AI can fulfil, and how the technology will influence the organisation as a whole.   As marketeers, we need to adopt the role of educators to keep our brands ahead. So work with whoever you need to craft and make a compelling argument. This is something I can help you to do, so please grab me at the end of the presentation if this is something you want to talk through.
• #56: Or if you have to scoot at the end of the session- drop us a message on Facebook and I’ll be more than happy to chat at a later date.
• #57: I’m not a coder, but I am a marketer and it’s my job to become an expert on this. As is it yours. In 2 years time – 95% of us will be using some type of AI-enhacement to our work In 10 years time, could be as important as reporting metrics for social In 20 years time, job roles will be incredibly different. Don’t know if the term AI expert will exist as it becomes more of our every day – a lot like digital marketing manager In terms of General AI? The experts don’t agree when it will happen, so I couldn’t possibly say either.
• #58: I’ll instead end on the words of the great Kevin Kelly, Executive Editor of WIRED magazine… We don’t know where exactly AI will end up in our daily lives, but we do know it’s going to become a bigger and bigger part of it.