August 23, 2017
5 minute read
Think of AI and Hollywood depictions such as those in I-Robot, Westworld or Her may come to mind. But what many may not realise is that AI is already being used in several recognisable forms today. What’s more, its applications for the commercial world are predicted by many digital experts to be the next technological wave, following in the wake of the mobile evolution.
Take the rise of the voice recognition-based virtual assistant. Apple’s Siri, Google’s Assistant, Microsoft’s Cortana and Amazon’s Alexa are all obvious examples of AI being used to enhance customer experience. Less apparent may be the AI behind the predictive preference technology used by companies like Amazon and Netflix to boost market share, which analyses user behaviour to recommend other products and viewing choices likely to appeal.
Essentially, AI helps to shorten the distance between a need for information and the information itself, benefitting both consumers and marketers. For marketers, not only does this represent an opportunity to use the technology to better serve their customers, it can mean a greater level of insight into consumer activity, as AI starts to enhance marketing automation itself.
One of the fastest-developing applications of AI is image recognition. Using ‘deep learning’ techniques, programmes capable of image recognition can learn to identify what’s in a picture by drawing on vast image databases. It’s one of the reasons internet giants like Google and Facebook encourage their users to upload images; these user-tagged pictures help to hone the abilities of their deep learning networks.
Earlier this year, Google revealed an upcoming image recognition app – Google Lens. The app will enable users to simply point their device (smartphone or tablet etc) at an object and have Lens identify it and find information about it, without any typed or spoken search queries.
Furthermore, Lens can instantly grab useful data about the objects it views. At the app demonstration in May 2017, Lens was pointed at a Wi-Fi router and immediately connected to it by using optical character recognition to read the username and password. As Lens can recognise business premises such as shops and restaurants and find relevant information about them at a glance, the marketing potential of such an app is huge.
Pinterest is another company using image recognition, so far to enhance its marketing platform. The image sharing social network is to introduce a feature that automatically suggests promoted pins based on what’s in the image that a user has pinned themselves. Pinterest is already using this technology for organic pins and will roll this out for paid pins too, picking up key characteristics in selected pins and showing advertiser pins that contain similar content.
At Quant, we have used image recognition techniques in a similar way to deliver a tailored solution to our clients. Our Style Puzzle is a gamified survey tool, whereby customers will select the images which they are most attracted to. By taking shape, texture, colour and pattern into account, this image recognition software captures a customer’s visual perception of products, allowing us to deliver recommendations and personalise further communications for a unique shopping experience.
While image recognition AI learns to derive meaning from pictures, devices such as voice recognition-based personal assistants used a form of AI known as conversational user interface (UI). The aim of conversational UI is to mimic chatting with a human, and many chatbots respond to both voice and text commands.
Applications that use conversational UI employ input recognisers and decoders to break down a spoken or written query, then apply a type of AI known as ‘natural language processing’. NLP uses learned knowledge of semantics to deduce the meaning of the query before the programme finds the required information and generates a ‘human-sounding’ reply.
Virtual assistants like Alexa adapt to user speech patterns, vocabulary habits and even accents, so the more you speak to them, the better they get at understanding you and the more personalised their answers can become. You can ask for a wide range of real-time information, such as local weather, directions and of course, shopping recommendations. Most virtual assistants also integrate with other apps too, so you can ask Alexa to book you a taxi through Uber, for example.
This instant access to multiple systems greatly streamlines user experience – instead of having to search numerous apps or websites for the information you need, a virtual assistant locates and relates the answer to your query in seconds. According to Think with Google, over 20 percent of searches are now voice-based, so the technology is certainly taking off, meaning that marketers have to be aware of how they can capitalise on this new communication platform.
But conversational UI isn’t all about voice. Other companies, such as Microsoft, are using this kind of AI to provide not only a more convenient consumer experience through the medium of the chatbot.
In 2016, Microsoft launched bots into the Skype app, which are designed to help users book, buy and stay up to date with all manner of products and services simply via chat. Available in the Contacts section of the app, users can browse bots from brands including StubHub, Skyscanner and UPS and complete entire transactions by instant message. Bot conversations are saved in the Chat section of the app, encouraging users to come back to them time and time again, as they would any chat with a friend or family member. And because the bots can access other systems on a user’s device, they can anticipate calendar clashes when booking events or appointments, or even proactively suggest nearby hotels or meeting up with friends who live in the area.
This emerging technology means that marketers may be able to incorporate AI into their customer-facing interfaces in ways they may not have realised. A case in point, Microsoft have opened up their bot data in their Bot Framework, which enables developers to create their own chatbots based on their coding. It’s now easier than ever to include an AI algorithm into your company’s software.
We’re seeing the emergence of AI generated content, made possible by algorithms that can identify key themes and phrases from a dataset and weave them into a written article that reads like it was created by a human author. Natural language generation (NLG) engine Wordsmith is already producing content for companies like Microsoft and Yahoo! which suggests others will follow suit.
Propensity models can be created by feeding huge sets of data into machine-learning algorithms which, for example, can be used to predict likelihood that an individual will make a purchase and determine how much they will be willing to pay. Through these models, AI can automate factors integral to marketing, such as determining the efficacy of leads, creation of a hyper-personalised customer journey and even the implementation of dynamic pricing to appeal to targeted consumers.
Mike Kaput of the Marketing AI Institute explains, “As AI is baked into marketing automation and CRM platforms, marketers may use the technology without even realising it. It will be the new normal. Eventually, the term “intelligent machines” will be redundant. All machine systems will use some type of artificial intelligence to augment human activities,” (marketingaiinstitute.com).
As part of their survey into AI, US-based ABM company Demandbase found that 80 percent of marketing executives believe AI will revolutionise marketing by 2020. It’s a very exciting time for marketers and those looking to unlock big data, and we can’t wait to see what’s to come. For now, talk to us today to start using data to get more from your marketing.