Delving deeper into VR, AR and AI: Six emerging themes

They have naturally and rightly become topics of concern beyond the auditoriums of Silicon Valley and it’s no surprise that marketing and advertising are talking the talk more than most. At Cannes this year, the relatively new but now prominent Innovation Festival was dominated by talks covering VR and AI. An opening slide of a robot or someone wearing a VR headset was de rigueur at the Innovation Festival, and it bled into the main festival too, with plenty of examples and insights into VR and 360 in particular.

These topics have been bubbling up in the ad world for the last few years, and have now reached fever pitch. Delving deeper beyond the acronyms, these macro trends are spawning particular problems, opportunities, and moments of progress – where the really interesting stuff lies. In my view, there are six topics to take note of, for now, next year, and perhaps beyond.

In this first piece I cover the first four, all of which are related to AI. Look out for a blog post with the final two next week.

Man wrestles the machine

“Will I be replaced by a robot?” the man quivered and typed into Google. With that attitude? Probably.

The fear of AI is real. Organisations and publishers continue to tackle this question, creating hitlists of the industries most vulnerable to machine replacing man. Inevitably, those who complete repetitive, logical tasks are most at threat, whilst us creative types in the ad world can smugly continue our sugar-fuelled brainstorms and booze-inspired ideas sessions, content in the knowledge that no robot would have the imagination to think up the next Cannes Lions Grand Prix winner – awarded by humans, of course. At this year’s Cannes, the man vs machine debate was rife, with a palpable tinge of fear. Talks included, ‘The Rise of the Robots’, ‘Artisan vs Algorithm’, and, cutting to the chase, ‘Will a robot win a lion?’

For me, this debate misses the point slightly. It shouldn’t be a question of man or machine. In reality man is creating the machine: the brains of humans are building systems that can then do their own ‘thinking’. We can determine what we want these machines to do and ultimately, their output. A machine can be given a challenge to tackle autonomously but unless you hook the machine up to the 'on' switch, man decides whether to put that into practice. In reality, the answer to the question of “Who will do x in the future?” is probably the less fantastical, but no less fantastic answer of ‘both’.

AI: quicker, better, or different?

AI is shrouded in mystery and blurred with science fiction, probably more so than any other trend in technology. Sometimes it is over-complicated and at other times over-simplified. For brands the real question to ask of AI is: what do I want from it? Do I want something to be quicker, better, or just different? Much of the focus on AI in the mainstream is on efficiency. Performing tasks 24/7, 365 days a year, with no lapse for a loo break or even a scroll through Facebook. This, of course, is a huge arena for AI, but one that I would argue is less the domain of marketing and more that of operations.

Where it gets more interesting for those not simply looking to save some money, is in the ‘better’. Can a machine provide better service than a person? Not quicker, not just more reliable, but more informed, personalised, and ultimately, enjoyable. This is where brands can consider user experience — not just at pain points but throughout the entire user journey — and determine whether AI could power their offering.

Then there is ‘different’, which should lead to better, but doesn’t necessarily (an important point in its own right). Google's DeepMind AI beat the world's top Go players by taking what expert human players deemed a bizarre strategy, powered by the rather haphazard but ultimately successful approach of a certain strand of machine learning. This is the most popular approach in AI circles at the moment and the one that could lead to answers that our minds – ‘clouded’ by standard practice, logic, and tradition – may never find in a million years.

Bots as the new apps

AI-powered customer service is something that has been spoken about for some years. In December 2013 I led a project for Panasonic that created an AI-powered customer service channel for customers in department stores. When in-store and pondering whether the Panasonic product they were considering could do x, they could text a shortcode to ask any question directly to the brand. Unbeknown to those customers, the text conversation they were having – and many did, with an average of 10 messages per user – was not with a person, but software. This was a successful project, but a hard sell to brands: “Let a machine represent your beloved brand!”

Fast-forward three years and AI-powered customer service has a far catchier, friendlier moniker: chatbots. And they’re big news. More than 11,000 bots have already been created for Facebook Messenger – the most popular platform to house your bot. For me this highlights two points of interest. For the first, see the previous topic: do you want your bot to be quicker, better or different to your existing digital comms platform? The answer should probably be all three.

The second is a matter of discovery. In this sense, bots feel like the new apps. Thousands of developers are independently creating bots and pushing them out into the digital ether, but there is no clear single way for people to discover them. This ecosystem will certainly mature, but the challenges brands have had with app discovery looks set to be echoed with bots, and lessons from apps created at great cost but downloaded by few should be heeded.

The comparison of bots to apps has a far more positive spin too. Apps created by brands should be useful or entertaining for customers on a regular basis over a significant period of time. Indeed, the aforementioned failure of many brand apps is because it did not tick these fundamental boxes rather than it being a bad customer experience. If a bot, as with an app, is created with thought and for the right reasons then it can reach a marketing holy grail – consistent, regular dialogue with customers that leads to a relationship of mutual understanding and appreciation.

Understanding language

To expect machines to ‘play human’ there is one fundamental challenge before you even get to the part where AI attempts to replicate the astounding thought processes of the brain. That challenge is language. Any machine aiming to understand what a human is asking of it or provide a meaningful output truly needs to understand language, or any level of super smart ‘working out’ is lost. Life is not a game of Go.

In my experience of AI to date, from chatbots to new forms of search, it’s the lack of understanding of the beautiful subtleties and nuance of language that has led to a disappointing experience. In customer service specifically – the most obvious example of this – the last two decades has seen corporations outsourcing customer service to call centres abroad, before bringing them back to the UK due to customer complaints of difficulties in communication and service assistants uttering ‘robot-like’ set lines from the screen in front of them. Brands must remember this experience when creating this type of chatbot.

It is no surprise then that a whole sub-section of the AI scene is companies looking to crack comprehension of languages. Perhaps it is indicative of the enormity of this challenge that most Facebook Messenger bots pre-determine answers for the user, with multiple-choice text responses or a scrolling carousel of options, rather than allowing a free-flowing conversation with all the complexity and ambiguity that brings.

Sam Battams is head of tech innovation at Manning Gottlieb OMD. Part two of this blog will run on thedrum.com next week.