Why AI needs a magic moment

Well-known author Malcolm Gladwell defines the tipping point as “that magic moment when an idea, trend, or social behavior crosses a threshold, tips, and spreads like wildfire.”

The idea of using AI-powered technology to fuel digital marketing and customer experience has been around for years, yet recently it seems that every leading newspaper and tech blog is buzzing with AI stories. Major analyst firms too have been predicting an AI investment boom. IDC, for example, is forecasting a 54 percent growth in marketing spend on AI software over the next four years, from $360 million in 2016 to more than $2 billion in 2020. We’ve also seen an uptick in deployments of AI-powered machine learning offerings from marketing cloud vendors racing to provide the optimal automated experience delivery.

AI has the potential to help make sense of the vast amount of behavioral data and customer signals that are difficult to pinpoint amidst much noise. It can also scale customer interactions to get to the elusive one-to-one interactions that all marketers strive to attain. However, it has to be applied to meaningful customer data and meaningful personalization tasks, or the results will continue be meaningless.

So, is the recent craze over AI just hype? Or have we truly entered a new era of AI-led digital transformation?

Current state of AI innovation

First, let’s take a look at the technology itself and whether recent breakthroughs might be contributing to the rise of AI.

The most recent and exciting class of algorithms is referred to as “deep learning.” These algorithms facilitate Artificial Narrow Intelligence (ANI), which is the lowest caliber of artificial intelligence. With the recent trend of employing deep learning through web technology, we can get machines to outperform human-level performance when it comes to speed, precision, and scalability. Tasks that take a human less than 30 seconds — yet require the analysis of data that’s voluminous (Google-scale) with underlying patterns — are ideal for robots. Face recognition, speech recognition, lip-reading, and similar activities rooted in analyzing images or time series are good examples.

Additionally, the combination of advances in deep learning and the implementation of experimental frameworks in simulated environments is leading to advances in the field of “deep reinforcement learning.” Consequently, we are likely to see an explosion in simple-to-automate repetitive manual tasks, such as those performed by self-driving cars and warehouse-packing robots. Before you know it, robots will be stacking shelves, bagging your groceries, and handling your airline’s check-in luggage.

The impact of AI innovation on the digital customer experience is likely to be felt primarily in the mechanisms that companies use to interact with their customers. The potential for time savings is already impressive — for instance, our team uses machine learning for tasks such as uncovering underperforming customer segments that a brand can influence. Every 10 underperforming segments unearthed automatically through the machine learning engine would take four and a half employees working full-time for a month to accomplish. Another area of interest and growth in terms of practical applications of AI is conversational commerce, which refers to the intersection of messaging apps and shopping.

Data flood yields marketing performance gap

While AI-powered technology continues to advance, the marketer’s ability to leverage the data being produced by every aspect of the growing marketing stack continues to deteriorate. One major reason is the sheer overabundance and constant flow of data. The volume is so enormous that we are seeing human decision-making become inefficient and inaccurate. Then there’s the ability to pinpoint actionable insights. Think diminishing returns — initially, we can implement easy changes, but after that you have to dive deeper and deeper into the vast sea of data. This means gleaning information and identifying patterns becomes increasingly more challenging for the human brain. At some point, the value of the insights fails to increase proportionately with added investment.

In a recent paper published by IDC and Qubit, analysts Gerry Brown and Philip Carnelley explain that at present we are seeing a marketing performance gap. Specifically, traditional B2C website marketing methods have started to flatline, and there is now little room for further improvement in their performance. A/B testing in particular — an experiment where users are shown two or more variants of a webpage and improvements are made based on feedback — is no longer achieving the performance required by B2C businesses.

Additionally, a flatlining effect is evident in traditional customer analytics tools used for analysis and reporting of historical data. Predictive customer analytics is having little impact on marketing performance for those wanting to go beyond traditional data mining, since marketers have been unable to convert analytic insights into dollars.

The decreasing quality of customer data collection is actually having a negative impact on many businesses, leading to a “marketing performance gap” between customer analytics and monetizable marketing actions.

How AI strategies can help brands stay competitive

AI-powered technology has come a long way. It can address the growing marketing performance gap. It can also significantly enhance customer engagement by turning a company into a tangible personality — a “living” brand in a more real sense than it is now. For brands, this means a heightened ability to respond and engage with customers in real time. For shoppers, it’s about identifying with the brand in a more personal way — not just making a purchase, but rather deciding to “join” a brand based on the customer experience.

Executives needn’t fear this transformation. Instead, they can embrace it by educating themselves on the fundamental applications and limitations of AI. This will help them evaluate the areas of their business where AI-fueled technology will be most impactful. They can start by identifying how they plan to use AI by asking themselves: Are we clear on the problem we are trying to solve? What parts of the usual workflows will AI touch? And how will human teams across the organization support AI?

Then, they can assess the technology for fit. Take customer experience delivery, for example. Is there a seamless integration between the back-end systems and store-front and experience delivery mechanisms? Is there a rapid feedback system that facilitates iterative interaction from dialogue to collection to insights to actioning and measurement? What about metrics of customer engagement? What are the best processes to test and monitor the ability of these AI technologies to add value and liaise with human operators?

Finally, they must remember to invest in an off-switch. Sometimes things go wrong, and investing in sufficient redundancy to minimize the impact of a failing AI technology is important. Challenges with Microsoft’s chatbots, Google’s misclassifying of automated face recognition systems, and Knight Capital’s flash crashes come to mind.

It’s all about humans and robots coexisting

So, are we at a point where we have to choose between machines and humans? No, and here’s why. It would seem that — thanks to sheer efficiency — AI-powered technology would win every time. But on its own, AI is still inferior when it comes to abstract reasoning and instinctual understanding. If provided with the wrong data or tools, the untrained artificial brain will fail. Robots need human guidance. As AI technology continues to mature, humans will become increasingly more focused on tasks that exploit human strengths, like the ability to detect patterns on an intangible level and formulate theories about the nature of objects and ideas.

We should not be approaching this new era of AI as a war between human and machine. Instead, we need to think human plus machine and identify how to create the most synergistic combination of strengths. There’s a huge opportunity to free up human time and focus it on the many areas that require the human touch. Only when we master this combination of strengths will we see a true AI-powered transformation in digital marketing.

Graham Cooke is the founder and chief executive officer at Qubit, the pioneer in data-first customer experiences.

Above: The Machine Intelligence Landscape This article is part of our Artificial Intelligence series. You can download a high-resolution version of the landscape featuring 288 companies by clicking the image.