From the future of work to the subversion of democracy, 2017 has seen artificial intelligence subject to a wide variety of associations. Stemming from scientific breakthroughs and heightened by the allure of existential risk, the implications of AI can seem grave if not overhyped, with a tendency to engender more confusion than strategic clarity among marketers. In truth, there are very exciting things going on in AI research. Coupled with and enabled by our data-driven economy, the advancement and growing ubiquity of AI software stands to dramatically impact the global economy in coming years.
But marketers should be careful not to conflate stories about experiments in deep neural network technology (such as headlines about AI’s ability to master our most challenging strategy games on their own, predict your sexual preference, and invent its own language) with game-changing technologies poised to dissolve our professions (much less our progeny). Working in a creative industry, marketers are well-positioned to thrive in an AI-enabled environment, particularly in the short term.
Think of AI as a collection of software techniques used to make computers reason similar to how humans reason. Machine learning is a branch of AI techniques that entail configuring an algorithm to automatically adjust itself based on large amounts of data. Plenty of machine learning methods exist (mostly amounting to robust statistical procedures), but neural networks are behind the most jarring headlines you’ll come across. Neural networks are somewhat mysterious algorithms created by processing extremely large amounts of data through synaptic structures, similar to how a human brain functions.
In the context of human capabilities, today’s AI can only perform a relatively small range of skills – it’s so-called narrow or weak. The holy grail for AI research is general or strong AI, a system that can rival human intelligence in sufficient domains so as to be considered convincingly human. With data comprising the backbone of AI and neural networks showing so much promise, an arms race is underway among researchers in government, academia, and the private sector seeking to master AI. By training their neural networks with enormous amounts of data, these researchers use a process known as deep learning to discover nuanced patterns that human cognition is incapable of reaching alone.
The thing is, though, these sorts of technologies are already commonplace in consumer tech. For instance, both Google Search and Google Translate are technically powered by AI, as are Facebook’s news feed and Spotify’s Discover Weekly. Because it fundamentally entails large amounts of image data, products involving computer vision such as Snapchat’s facial recognition, Google visual search, and Pinterest Lens also necessarily rely on AI. Autonomous technology also forms the technical basis for programmatic advertising itself – perhaps unsurprising as ad exchanges are rooted in trade automation. In truth, the technology is all around us as marketers; 2018 will just be the year when it most noticeably finds its way into more novel marketing solutions.
When reckoning with speculations about AI and the future of work, it’s helpful to think about human capabilities in four main skill categories: manual routine, cognitive routine, manual nonroutine, and cognitive nonroutine. In theory, manual routine tasks found in places like factory working and assembly lines will be the quickest to automate, whereas cognitive nonroutine tasks such as being creative (or any job entailing a high degree of interpersonal interaction, such as nurses or social workers) will grow in demand. This is because nonroutine cognitive tasks are inherently much more difficult to quantify, and new technology always entails a degree of technological unemployment.
There remains a vibrant discussion about whether job creation will outpace technological unemployment; because most marketing activities fall into the savory quadrant of cognitive nonroutine tasks, our profession stands to be largely insulated from complete job automation for the foreseeable future. But let’s not kid ourselves into thinking cognitive routine tasks are not a loathsome aspect of modern marketing. Anyone working in analytics, media, or traditional digital marketing such as search engine optimization (SEO) or search engine marketing (SEM) at a large agency knows the countless hours of routine number crunching and other cognitive chores are endemic to the attention economy – thankfully, this is where AI stands to most significantly benefit marketers in 2018.
Just as AI is not new to consumer tech, it is also not new to marketing. Most existing solutions derive their value from customer relationship management solutions, like Salesforce’s Einstein and Marketo that increasingly bake in machine learning functionality that allows for predictive and anticipatory lead-generation tactics. Adobe Smart Tags and Google’s Cloud Vision API also allow for smart digital asset management services, affording the automation of display ad trafficking by using computer vision to generate naming conventions on the fly.
Companies such as Affectiva and GumGum offer a glimpse at the emergent computer vision space, using facial recognition to track expressions and offer so-called “emotion as a service” for novel testing methodologies, as well as application programming interfaces (APIs) for creative experimentation. For retailers, providers such as ShopperTrak and RetailNext use computer vision to learn more about customers, leveraging continuous footage of shoppers inside and outside their stores to serve more relevant offers and advertising. Vendors such as Clarifai and LogoGrab use similar tech to provide services along the lines of “visual listening,” scraping the visual Web for instances of your brand or desirable user-generated content (UGC), as do existing social listening and sentiment analysis platforms such as Crimson Hexagon and Synthesio.
Even creative roles are in the sights of AI technologies. As we produce more data, neural networks will also grow more capable, increasingly able to augment routine aspects of creative work. Manifest in software like Adobe Sensei, Grid.io, and Wibbitz, these proficiencies are still budding but show great promise. And while programs that automate copywriting tasks, such as Persado and Automated Insights, can be useful for smaller-budget campaigns and copy testing, these technologies are far from diluting creative capabilities. Rather, they stand to enhance artistic faculties, stimulating and augmenting the creative process by analyzing creative work and iterating against aesthetics, archetypes, or emotional nuances.
As routine cognitive tasks take up decreasing mindshare among marketers, nonroutine cognitive activities grow in importance. In this environment, AI’s role in creative activations and tactical approaches is tantamount to its role in operations technology. Some of the most powerful brand activations moving forward will also be made possible with the emergence of new AI technologies. Following are three activation paradigms bolstered by the rise of machine learning and neural networks.
Alongside AI and social platforms, the crowd is said to be a pillar of the new business playbook. It follows that marketers should orient themselves toward engaging the crowd for more innovative campaigns using AI. This will not only afford cultural benefits when well executed, but can also be used as a means through which to acquire novel forms of customer data.
Though it’s unclear whether data was collected to improve targeting efforts or contribute to a larger campaign, the Snickers Hungerithm is an example of leveraging crowd culture with AI to great effect. The Hungerithm was essentially a crowdsourced algorithm based on the collective sentiment of the Web, dictating the price of Snickers at participating 7-Elevens, with the idea being that cheaper Snickers bars would help satisfy our collective hangriness. Because the rise of social and mobile has largely led to our data deluge, it makes sense that the intersection of both is fertile ground for brands to leverage the crowd. This is particularly true when an interactive experience tied to the physical world is calibrated using AI in real time, such as in the case of Snickers’ Hungerithm. More interestingly, brands are charged to experiment using this data to better inform planning about participating customers, working toward building a robust micro-influencer ecosystem for future campaigns.
Lean Into Experiential
AI affords novel opportunities for brands to interact with users in the real world and allows novel types of legitimate data collection. Not only are experiential activations notable for their capacity to familiarize society with AI, they also allow for novel types of legitimate data collection, with AI-based monitoring capabilities improving measurement and attribution modeling. Individual consumer data can be analyzed to better track marketing dollars, with technologies such as visual listening, expression tracking, and sentiment analysis allowing marketers to better calculate impressions and engagement (provided the experience is shareable, of course).
An early example of this was Chevrolet’s Positivity Pump, which let people receive discounts on gas based on the positivity of their social media profiles. Similar to Snickers’ Hungerithm, this activation centered on the negativity on social media – a cultural tension interestingly related to AI-enabled social platforms. The activation used IBM Watson’s AI technology to do something novel and legitimately valuable with social data. With experiential campaigns, be smart: The best experiential activations don’t turn consumer content into brand schlock; they respect their consumer by ensuring their activations are things consumers want to share, not what they’re forced to share.
Raise Your Chatbots
Facilitating UGC contests with chatbots is one way to leverage the shortening relationship between consumers and brands. The most interesting examples of chatbot activations involve their serving as input vehicles for AI-powered personalization engines.
For instance, executions such as Sephora’s Virtual Artist chatbot used a computer vision API to detect faces in user-submitted photos – paired with smart technology on the back-end, this activation essentially allows people to try on makeup before they buy it. Other brands such as Perrier and Denny’s have used chatbots to customize photos and deliver coupons through choose-your-own-adventure campaigns. With any chatbot strategy, data collection protocols are fundamental. It is imperative to create the most useful and interactive chatbot experiences, building relationships in ways that both elevate your brand and garner valuable consumer data for future campaigns.
In 2018, some marketing executives will be tempted to view AI investments as simply a way to cut labor costs. As business leaders, marketers have a responsibility to retrain, reinvest, and even consider readjusting revenue models and talent schemes to leverage AI in creative new ways, as opposed to simply reaping marginal efficiency gains to stay competitive in the short term.
On personal and professional levels, marketers will be increasingly reliant on AI-driven algorithms for day-to-day information gathering. Over time, this algorithmic deference will transform our cognitive abilities; our minds and modes of thinking will continue evolving alongside technology. But in overrelying on information feeds (as opposed to seeking new information, media, and aesthetics for ourselves), we risk compromising our mental gatekeeping. Assessing this scenario, the rise of AI challenges marketers to be vigilant in fighting algorithmic overreliance, pushing our appreciation for experimentation, divergent thinking, and novelty in all aspects of life. In realizing this opportunity, we can use AI to enhance our perspective, reinvigorate cultural vibrancy, and, by way of action, fight the complacency endemic to our current condition.