You didn’t have to be following the coverage from this year’s CES too closely to notice one name repeatedly cropping up, across what felt like every announcement at the show: Alexa.
Alexa is an artificial intelligence, a virtual personal assistant developed by Amazon. And right now, it looks likely to be remembered as AI’s first major foray into the mainstream.
Developed by Amazon’s Lab126 R&D division, and building on the eCommerce giant’s acquisition of Cambridge AI startup Evi in 2012, Alexa made its debut – there’s a temptation to say ‘her debut’, given the gendered name and voice, but let’s stay neutral for the moment – back in November 2014. It was first sold as part of the Amazon Echo, a smart speaker that enabled users to ask questions, select and listen to music and, of course, order shopping using just their voice.
It wasn’t until last year, however, that Alexa made a real splash.
Amazon reported that the Echo, along with the smaller Echo Dot, were its best-selling products of Christmas 2016, selling nine times as many as the previous year. While the retailer doesn’t release any specific sales figures for its products, Morgan Stanley estimates that 11m Echo units had been sold by the beginning of December – a figure that, notably, doesn’t include the Christmas rush.
While Alexa is currently leading the conversation around AI assistants, however, it is far from the only one in market today. Google has Assistant, which followed in Alexa’s footsteps in late 2016 by launching first with a smart speaker, Google’s Home device, before making the move to Pixel smartphones and the Allo messenger app.
Meanwhile, Facebook is working on M, an assistant for Messenger which started initial trials in August 2015 but isn’t expected to be widely available for “years”, VP of Messenger David Marcus said at F8 the following year. Microsoft has Cortana, pitched as the heart of Windows 10 on both desktop and mobile, as well as the Xbox One games console, and Samsung is prepping its own assistant, Bixby, after acquiring Viv – a company established by former creators of Siri – last October.
The matriarch at the top of this ever-growing family tree is, of course, Siri. Introduced by Apple back in 2011, Siri may be a little basic compared to some of the AI technology developed since, but it laid the groundwork for everything that followed. Apple introduced to the mainstream the idea of a voice that can respond intelligently to your questions, and Siri entered the public consciousness more fully than any assistant since, making appearances in everything from TV sitcom The Big Bang Theory to Marvel blockbuster Ant-Man. It also served as the inspiration for films like Her and Ex Machina, which explore human relationships with female-presenting AI – stories that are looking increasingly prescient, as Amazon reports Alexa has received over 250,000 marriage proposals.
All of these assistants are varied in terms of what they can do and what platforms they’re available on, but they have a few things in common. With the exception of Facebook M, they all respond to the user’s voice, using natural language processing to interpret their requests, and tie together a number of services – first- and third-party – to achieve their given goal.
What made Alexa’s dominance of CES this year so notable was that Amazon itself was not present at the show. Instead, the announcements all focused on the assistant’s integration into other devices, from its first smartphone, the Huawei Mate 9, to a range of ovens, fridges, washers and dryers from Whirlpool. The list also included Ford’s Sync 3 in-car system, Ubtech’s Lynx robot, and Aristotle, a child-friendly version of the Echo from toy brand Mattel.
Rise of the machines
When it comes to AI, these kinds of personal assistants are just the tip of the iceberg, as a quick look at the headlines of 2016 will attest.
Facebook CEO Mark Zuckerberg kicked off the year by announcing his project for the next 12 months: “to build a simple AI to run my home and help me with my work”. In December he showed off the result, nicknamed ‘Jarvis’ after the AI butler in Marvel’s Iron Man films. Controlled using a mix of text and voice, via Facebook Messenger or a custom-built iPhone app, Jarvis is able to switch lights on and off, open the house’s front gate and even make toast. It’s essentially a personalised version of Alexa, with the added bonus of being voiced by actor Morgan Freeman.
Facebook, Microsoft, Slack and Kik all introduced chatbot offerings, enabling users to have text conversations with basic AI bots through their messaging platforms. This came just weeks after Microsoft’s ‘Tay’, a Twitter bot designed to learn from its interactions with users, had to be shut down 18 hours after launch when the bot, having apparently learned a little too much from its time on the internet, started to post racist, sexist and otherwise offensive tweets. Not to be deterred, by the end of the year Microsoft had revealed Tay’s successor: Zo, a chatbot that tried to address previous mistakes with a limited launch, on messenger app Kik, and a restriction on discussing sensitive topics such as politics.
Meanwhile, Google’s DeepMind was busy beating the world number one at the ancient Chinese board game of Go, before moving on to its next competition – the much more modern game of Starcraft II – and partnering with the NHS. At the end of March, ad agency McCann Erickson even reported its first non-human hire, in the form of AI-CD, an AI creative director that mines its database of existing TV commercials to provide ‘logic-based’ direction for new ads.
It’s no surprise, then, that Google CEO Sundar Pichai said that the next 10 years would see his company, and the rest of us, “shift to a world that is AI-first”. Pichai described “a world where computing becomes universally available – be it at home, at work, in the car or on the go — and interacting with all of these surfaces becomes much more natural and intuitive, and above all, more intelligent”.
That process is already well under way. “AI is all around us already, even though we may not be aware of it,” says Thomas Husson, VP and principal analyst at Forrester Research. “AI techniques are in action whenever we use a voice-driven assistant like Alexa or Cortana, allow an online utility to tag our pictures, or when we receive recommendations about what we might want to watch. Progress has accelerated in the past few years, with well-publicised success stories such as Google’s DeepMind technologies beating a professional Go player or IBM Watson for Oncology supporting clinicians.”
So it looks like the rise of the machines is well and truly imminent. But, before we consider what happens next, it’s worth asking: how did we get to this point?
A brief history
“The term AI was coined at the Dartmouth conference in 1956,” says Husson. “Since then there have been a couple of key milestones, both in technological progress – IBM’s Deep Blue in 1997 and Watson in 2011 – and in people’s imagination, with movies like 2001: A Space Odyssey, Terminator and Steven Spielberg’s A.I.”
To date, as these milestones suggest, progress in this field has been primarily led by one company. “IBM has been researching, developing and investing in AI technology for more than 50 years,” says IBM Watson’s European leader for retail Tony Maile. “When IBM Watson won the Jeopardy! gameshow on TV in 2011, it was a watershed moment. Since then we’ve only accelerated our innovation, advancing and scaling the Watson platform and applying it to many industries, including healthcare, finance, commerce, education, security and the Internet of Things.”
IBM’s total dominance has been challenged in recent years, however, as a wide range of companies have begun to invest in AI, causing it to blossom in the myriad ways catalogued above. But does AI have a place in marketing?
“Brands cannot afford to ignore AI, as it forms the foundation around which we will architect the new customer experience,” says Matt Gee, head of digital transformation at Isobar. “AI empowers brands to truly put consumers at the centre of the experience, with contextually relevant brand touchpoints, curated for the individual and moment.”
A 2016 report from J. Walter Thompson found that 70 per cent of millennials in the US, and 62 per cent in the UK, said they would appreciate a brand or retailer using AI technology to show them more interesting products. And 72 per cent and 64 per cent, respectively, believed AI would be able to accurately predict what they want, as the technology develops.
All signs, then, point to AI becoming a vital tool in the marketer’s arsenal – but given that AI is a very broad church, where exactly should they be putting their budget?
Deus ex messenger
Arguably the most mature incarnation of AI marketing at this point is the humble chatbot, which handles both questions and answers in a messenger-style text format.
Chatbots have been around since at least the turn of the millennium, but have seen a major boost in the past 18 months, with just about every messaging platform you could name – including Facebook Messenger, Skype and Slack – introducing bots. At their most basic, chatbots are just a programmed set of text responses that trigger based on user inputs, offering everything from weather updates to text-based games.
“Chatbots enable an instantaneous dialogue with brands, which if deployed correctly can do the heavy lifting of customer service, leaving human interaction to resolve issues, thereby creating significant business efficiencies,” says Gee. “In a world where messaging has overtaken social networks, the new creative territories agencies need to focus on are personality, dialogue and conversational narrative.”
Burger King, H&M, 1-800-Flowers, and KLM are just some of the names to jump on board with bots, with the use case for brands dividing into two main fields. The first is automating CRM channels like call centres and online help, lightening the load on human staff.
“Traditionally, we saw customers getting support and access to websites from user interactions such as FAQ sections, live chat, email web forms and ‘call me back’ buttons, but the latest advances in AI point to a convergence of these channels,” says Jordi Torras, CEO of Inbenta. “Instead of all these different ways to access information, in future customers will require more and more actual conversations.
“Chatbots will quickly move from a ‘nice to have’ feature, as they have been in the last decade, to a central way of communicating with customers. I believe the industry will move from CRM to CCM (Customer Conversation Management).”
The second is what has become known as ‘conversational commerce’ – that is, placing an order for everything from a taxi to a taco without ever having to leave the chat window.
“Chatbots have proven useful at automating parts of the sales process when customers are looking for a self-service option,” says Giles House, CMO of CallidusCloud.
But it’s not a perfect solution, House says: “Chatbots can only handle the questions and responses they have been deemed most likely to receive. In any selling situation where the buyer is looking for specialised contextual information, chatbots are unlikely to deliver feedback decisive to closing the deal.
“People don’t buy from bots, unless the purchase is fairly basic, like a refill of some commodity goods. More complex sales need a human touch to establish trust, to provide answers to unique questions, and to apply human thought to the components of a solution. Chatbots may capture the raw data around those answers but they still can’t deliver on the human part of that equation; they can’t bring empathy, trust or a sense of teamwork into a deal.”
More than meets the eye
One way of introducing some of that human touch – in terms of presentation, if not actual content – is by communicating with users through voice, which is where AI assistants come in. The marketing use cases for assistants are currently very similar to those of chatbots, being split between basic CRM and commerce.
Examples of the former include EDF Energy, which produced a ‘skill’ (what Amazon calls the simple app-style prompts that users can add to Alexa) enabling customers to check their energy use or provide a meter reading; Capital One Banking, which did much the same for banking; and National Rail, which provides live updates on train times.
As for the latter – well, aside from the likes of Domino’s Pizza and Uber creating their own skills, it’s worth remembering that the Echo is an Amazon product, and one of the key selling points is the ability to buy things from its store with voice commands. Conversational commerce is likely to have a bright future on AI assistants – or, at very least, on the one built by the world’s biggest retailer.
There are opportunities for softer brand-building exercises on these platforms too. Tequila brand Patrón has an Alexa skill for cocktail recipes, while the food brand Hellmann’s offers a similar service for meals, and Johnnie Walker has a skill that guides users through a personalised tasting tour of its whiskies. It’s a brave new world for brands, which throughout the history of advertising have been mostly focused on the visual – something that is absent with a voice-driven interface.
“Brands need to figure out how and where they can appear within this context of AI assistants, in an authentic and useful way,” says Isobar’s Gee.
Behind the curtain
It’s also worth remembering that not all AI in marketing is consumer-facing.
“Beyond chatbots and assistants, the impact of AI on marketing is huge,” says Forrester’s Husson. “AI will drive an insights revolution that will help augment and enhance interactions between brands and consumers. Executives have told us that today, marketing and sales is the main function where organisations are evaluating investment and adoption in AI systems.”
The algorithm rules supreme in marketing today. Speak to anyone about AI’s role in marketing and the conversation is likely to turn at some point towards a seemingly unlikely topic: House of Cards, the award-winning Netflix drama series.
“All the films and drama series provided by Netflix are tagged. The database is connected to behavioural data, which they had been accumulating over six years. As a result, Netflix became confident that they had a full grasp of the user preference,” says Shun Matsuzaka, creative planner at McCann Japan, explaining the inspiration for AI-CD.
“Therefore, rather than purchasing expensive external content, they thought that they would be able to create better content themselves. House of Cards was produced based on an algorithm which concluded, using this database, that a political show starring Kevin Spacey and directed by David Fincher would be a major hit. Even though it was their first original series, Netflix invested $100m – they were that confident – and the show went onto win multiple Emmy awards.”
Meanwhile, the rise of programmatic is a major step towards an AI-led advertising landscape – and not just on the buying side either. As well as McCann’s AI creative director, last year saw the first machine-created movie trailer, when IBM’s Watson cut a trailer for, appropriately enough, robot slasher film Morgan.
“Watson is helping marketers gain a deeper understanding of consumers’ needs and wants,” says IBM’s Maile. “80 per cent of the world’s information is comprised of dark data: text, tweets, blogs, articles, videos and sounds that traditional programmatic computing can’t understand. With cognitive computing, we can augment the marketing professional’s ability to understand, and act upon, this complex data in ways that simply were not possible before.”
Of course, the obvious question when discussing AI is where it’s headed next. But before we start considering whether the future is more Terminator or Her, it’s worth noting that the technology still has some daunting obstacles to overcome. “AI is a hugely transformative technology,” says Forrester’s Husson. “It will have strong legal and ethical implications.”
For a start, all of the problems we face today on mobile with security and privacy will not only be inherited but amplified, given the intimate relationships that AI is trying to build. This has already started to manifest – as in the 2015 murder case where the prime suspect owned an Echo, leading to US police issuing a warrant for Amazon to hand over the device’s audio logs.
The company refused, but it was enough to unnerve other Echo users. There are plenty of videos online of people asking “Alexa, are you spying for the NSA?” – to which the device responds, naturally, by turning itself off.
The way that children will interact with AI devices is also a major concern. Another viral video starring Alexa, which did the rounds at the end of last year, showed a young boy asking Echo to play him a song, and being rewarded with a string of pornographic terms. That was followed by a news story in January about a six-year-old accidentally making a $200 order – which, when it appeared on a TV news show with the phrase “Alexa ordered me a dollhouse”, managed to trigger viewers’ own devices into making the same purchase.
“Brands will become more risk averse to any AI that interacts with children,” says David Skerrett, managing partner at Nimbletank. “There are some obvious trip wires around children using AI via voice devices.”
And this is just the start. As the technology evolves, we will have to face considerably bigger moral quandaries. If AI does become a vital part of the workforce, there is the problem of what happens to the people whose jobs are eliminated. Beyond that, if AI starts to achieve sentience, this raises the question of what legal rights the technology itself will have. It might sound far-fetched, but this is something the European Parliament has already started to consider, commissioning a report on whether AI will be considered ‘electronic persons’ and whether including a ‘kill switch’ should be mandatory practice.
But we’re getting ahead of ourselves. For the moment at least, you probably don’t need to worry about your Echo turning against you, and Facebook chatbots aren’t likely to be applying for personhood status any time soon. Even in the nearer term, though, there’s still plenty to be thinking about – so let’s leave the final word to our crystal-ball holders.
“Some of the key trends are going to be improvements in machine learning and natural language analysis capabilities,” says IBM’s Maile. “Personally, I am most excited about the advances in video analysis and the understanding of objects, human emotions and activity. We are only beginning to imagine what this means for providing relevant and exciting content to consumers when and where they want it.”
“A future where content is co-created with AI will definitely come,” says McCann’s Matsuzaka. “When it becomes the norm, creators will be leveraging data and algorithms in their creative development. In other words, data analysts and AI developers with a creative mind will be in high demand. Technology is evolving at an extreme speed that exceeds our imagination.”
“Voice is the future,” says Jorrit Van der Meulen, VP Amazon Devices and Alexa EU. “We believe that voice will fundamentally improve the way people will interact with technology. It can make the complex simple – it’s the most natural and convenient user interface, and the one we all use every day. It’s still early, but we think there’s a lot of potential in this space and we’re working hard to innovate quickly.”
“Facebook will compete with IBM, Apple and Google in terms of cost-effective access to AI for brands,” says Nimbletank’s Skerrett. “The UK’s top university degrees will become focused on AI. The technology will be used to create more content and robots in retail environments will become more normal.”
“The development of deep learning has ended an era of ‘AI winter’, which started decades ago, and created a huge wave of excitement,” says Inbenta’s Torras. “As powerful as it might seem, I believe deep learning is just the first of a series of algorithms that will be created in the future that will boost neural networks to learn to solve increasingly complex and abstract problems.
“Ultimately, I believe that AI will totally remove jobs that we today consider boring, freeing the humankind for a totally new era of creativity. I believe we are seeing a revolution with a similar impact that the Industrial Revolution had two centuries ago.”
“Such is the potential impact of AI that IBM, Google, Facebook, Amazon and Microsoft have joined forces to publish research under an open license in areas including collaboration between people and AI systems, and the trustworthiness of the technology,” says Isobar’s Gee. “The coexistence of people and machines will become intrinsically linked. AI will change what it means to be human, our relationships with our world and with each other.”
This article first appeared in the February 2017 print edition of Mobile Marketing. You can read the whole issue here.