Vox pop: the top AI and automation trends for 2019

The Drum has published a special one-off supplement dedicated to the use of artificial intelligence (AI) in the marketing world, and was also banging the drum loudly at The Drum Arms in Austin at SXSW talking AI and automation. There is little doubt that AI and automation are transforming businesses and the world that we live in. What that means is we also need to be focused on the ethical side, the human side of the story.

Industry insiders consider the top AI and automation trends for 2019.

The Drum Network asked its members: ”What do you think are the top AI and automation trends for 2019?”

Alex Hamilton, head of innovation, Isobar

In 2019, AI will play a role in redefining our relationship with technology and how we source information. UI designers and developers are in the process of creating new interface systems that are responsive to skills people use in their everyday lives. This means interfaces powered by voice, images, gestures and even the human brain. Advances in machine learning are improving computers’ ability to recognise human speech, allowing us to communicate along more natural lines. I only see voice-powered interfaces becoming more relevant as we become increasingly limited by the bandwidth in our fingers – a person can type 50 words per minute, or say 150 words per minute. This trend is evident in how search is changing. More people are using voice-powered devices to search for information, with some grand predictions stating that 50% of all searches will be via voice by 2020. The number of search queries launch by images – known as visual search – is also forecast to grow in 2019. The world’s biggest technology companies have all launched visual search tools. As better algorithms drive improvements in accuracy and relevancy, I expect these tools to post solid gains in visual search volume. This shift in how we interface with technology has relevance for brands and retailers as the majority of current touch points are text based. This shift away from a screen-dominated, point and click era will of course take time, but we’ll see the seeds of this trend planted in 2019.

Sarah Whitfield, CMO, Buzzoole

The right creator or influencer can make or break a campaign. They need to be authentic, have a real and engaged audience and align with a brands target demographic. This is the minimum bar set by most marketers and as you can imagine, it’s not always easy to achieve when purely reliant on manual/human selection. At Buzzoole, we use proprietary technology to match brands with the best Creators through NLU and visual recognition in order to determine their brand affinity. By putting this data together with first party performance data (engagement rates, impressions etc) our intelligent system keeps improving and thus is able to provide marketers with Influencers better suited to their campaigns. That isn't to say that automation will take over the management of Influencer Marketing in the future, but it will help provide better-performing solutions. A good example of this is in our work with Eloquii, a plus size women's fashion company and a still relatively small, but growing fashion space. Eloquii had been using Influencers for four to five years and had worked in a labour-intensive manual way i.e. it required a manual/visual review of all Influencers. Since Eloquii already had Influencers that they knew were plus-size, we were able to take these Influencers and retrain our AI. It required more than 20,000 images to do so, but we were able to do it with certainty and bring down the still necessary, but hugely reduced human element.

Daniel Todaro, managing director, Gekko

AI, in probably its simplest form as voice recognition technology, sits in millions of homes globally in various guises such as Google Assistant and Alexa and for Generation Z and Alpha is the norm. Speaking to an AI device is almost innate for these generations, whether it is through your Sky Q remote or any voice enabled device. Link this to search and the advertising medium becomes increasingly more relevant to marketing, shifting traditional ATL spend into the AI sphere as a subliminal form of media which is perceived as being helpful rather than as intrusive. Voice search enables more convenient and enhanced access not only to shopping but also to services such as restaurant bookings, delivery, train, cinema or theatre bookings when required or when prompted through smart AI advertising which entices you to explore more. Adverts, sponsorship and traditional ATL will encourage you to search and spend via AI, making transactions relatively seamless and hassle free. AI voice assistants will increasingly become an integral to every brands marketing budget. Having an AI assistant may mean you never need to leave your home, just as things were depicted in The Machine Stops by EM Forster which is becoming increasingly more a reality than fiction."

Steve Todd, technical director, Mashbo

In the marketing world there is so much buzz around the more shiny, engaging iterations of AI - driverless cars and robot chefs, image recognition and conversational AI in the form of chatbots and virtual assistants. However, working in a software development consultancy that uses AI to help improve businesses across the globe, the biggest trends we have seen so far in 2019 are less exciting, yet still completely transformational. We’ve seen a marked increase in businesses seeking solutions that streamline day-to-day processes, in particular around administrative tasks. For this AI offers process automation, where computers are used to deliver processes faster and more accurately than human counterparts, and rule-based systems, which can be used to create software that will provide an answer to a problem in place of a human expert. We believe these are the most useful - and most common - broad implementations of AI that we have right now. They have the potential to transform every sector, in businesses of every size. Yes, they will render certain types of jobs redundant, but in turn it will also create new and exciting roles in the field itself. What’s more, in those businesses already employing automation to take over day-to-day administrative and auditing tasks, we are seeing a trend that many did not expect. A need to take on morestaff. Take, for example, one of our own clients, a busy student lettings agency in Liverpool. Since partnering with us to create automated property management software, PropertyCloud, it has seen its portfolio grow. Freed from endless admin and mountains and paperwork, its team is focused on more enjoyable and profitable tasks. They are working shorter hours, have more ability to work flexibly and morale has been given a huge boost. The outcome? Happier, more productive teams getting better results and contributing to business growth.

Arnold Ma, CEO, Qumin

I still remember my first computer, at the age of 10 in 1996: it took about 30 minutes to download a single MP3 song. Fast-forward just 23 years, Gigabit internet is readily available to many homes across the UK, it’s almost 35,000 times faster than my “state of the art” 28.8k modem back in 1996.

Now, imagine that rate of progression on “intelligent agents” that have the potential to exponentially improve their ability to mimic “cognitive functions”. Should we be afraid to embrace these new developments in tech for fear of being outwitted and made redundant? I can empathise with critics in many sectors such as manufacturing, production, and so on, because these are mainly driven by “cognitive AI” – Analytical AI that uses past experience to inform future decisions, which is already widely available. The other two types, human-inspired AI and humanised AI, driven by emotional and social intelligence respectively, are what create the most controversy. Emotional AI is the understanding of human emotions as well as cognitive intelligence. Humanised AI adds social intelligence and self-awareness to the list.

We should focus on understanding how and where AI can help in certain industries. In marketing terms, AI is essentially “human learning abilities on steroids”, so think about the channels and tasks that are based on logic, rather than creativity. Channels such as Paid Search. However, without emotional and social intelligence, AI cannot be spontaneous or creative like humans. The human mind conjures up weird and wonderful things. We create and share things that evoke emotions, that we think others will find amusing, interesting, cute, sad. We have spent so much time trying to understand and keep up with data driven marketing and digital media, that we’ve forgotten what really matters: people, cultures and ideas.

Paul Vallois, managing director, nimbletank

The big trend in AI in 2019 will be its position within the business. We expect an increasing number of businesses to progress, at pace, through the AI maturity index. For the last few years we’ve seen a proliferation of AI on the periphery of business. It’s been the domain of innovation, data science or digital teams. The focus, rightly so, has been on developing use cases for pilots and experimentation but the results are often not fully socialised within the business. In 2019 we expect to see more businesses rolling out AI solutions in an increasingly connected and integrated way. More aligned to the business’s future growth strategy and increasingly powering their products and services. With this will come the need to take employees on a journey, explain the impacts and opportunities that AI will create – the tasks it will remove but the roles it will create. The need to imbed a culture of continuous learning within the business and the consideration that will need to be given to defining and sharing the organisation’s ethical approach to AI. Employees need to know that there is governance in place. With this increased level of maturity will come a more connected customer experience. Rather than just implementing the entry level, chatbot functionality, businesses will be exploring how highly contextual product personalisation, visual search, facial recognition and conversational AI can provide a much richer, multi-channel experience for the customer. We also expect AI to be further imbedded in products as opposed to acting as a supporting service to the product. The rise of ‘robo advisors’ managing our finances, personalisation engines curating our watch/play lists, automated healthcare professionals supporting our wellbeing and an increase in robot interaction in our everyday lives. At the same time, expect businesses to continue to automate back office services at scale, with repetitive and functional tasks being assessed and RPA (Robotic Process Automation) becoming common place. The one constant will be talent and the scarcity of it. As businesses look to scale these solutions there will be increasing pressure on the current talent pool and on business and government to address this.

Gareth Owen, managing director, Roast

I expect the upcoming AI and automation trends to be two-fold. Firstly, I expect the conversation surrounding AI to become more informed and less sensationalist. People are always keen to throw “AI” into their pitch patter because it conjures a cool image of something out of Terminator 2. This has to change. As this area becomes better understood, clients will eventually start calling their agency’s bluff and AI will be re-framed as a useful tool rather than some kind of magic marketing wand. Secondly, I think it’s going to get easier to merge different streams of AI in order to create something truly valuable. For example, merging Facial recognition, Optical Character Recognition, and Voice Recognition could help us develop reliable biometrics, potentially having huge ramifications for the world of Market Research. AI will only ever be as powerful as the thought that goes into its development and, subsequently, its deployment.

Jim Bowes, CEO, Manifesto

One term we’ll be hearing a lot more of in 2019 is Robotic Process Automation, or RPA, where software trained by human workers takes over mundane, repetitive tasks which involve transferring data from one system to another. It used to be that you’d have to very carefully specify a set of steps for the software to carry out to, say, copy loan application details from a scanned document into a database. But AI’s can learn just by watching a person do it a sufficient number of times, lowering the barrier to entry. The pace at which successful RPA implementations are proceeding is picking up and it’s expanding beyond the traditionally paper-based industries like banking, insurance and law, into the automation of many more office tasks. The leading RPA companies like BluePrism, UiPath and Thoughtonomy, already large companies, will become much better known as RPA goes mainstream. This year will also be the year where all the talk about AI as something that’s just over the horizon dies down and we start to see a lot more AI actually being put into practice. AI is about to get a lot less glamorous as conjecture over superintelligences and hypothetical debates about ethics gives way to discussions about platforms, frameworks and best practices. AI strategists will become much more common as businesses get to grips with the technicalities involved and the long, arduous process of putting their data into a format that can actually be used to power machine learning applications and predictive analytics.

Sam Watson, head of mobile, Brass

I think the key trend this year will be test and learn. Although everyone talks about AI, and I’m sure there are experts in house and at agencies alike, there will still be a resistance from brands to take the big jump in to it. Much like mobile development (the iPhone is 12 years old now) brands and companies still tread lightly into the unknown as they don’t know what it will bring in terms of cost or impact to the business. AI and Automation will be the same. Brands should be looking at what data they want to understand better that they can’t already, think about the data sets they have access to and then trial some small under the radar tests to see what data they get back. It’s not a magic pill that suits everyone so the real proof will be what happens this time next year!

Statistical machine learning is the most used commonly used and effective type of AI already used by marketers and it is set to grow in the future, as data collection, enabled by Internet infrastructure increases. More data will make it easier to improve the performance of this technology, that is to say that its performance and accuracy will be greatly improved. As the role of AI across many sectors has become crucial for informed decision making, it is interesting to look at what role it plays in an emerging industry such as Influencer Marketing.

Get The Drum Newsletter

Build your marketing knowledge by choosing from daily news bulletins or a weekly special.