Having the opportunity to work with comprehensive, boundless proprietary data assets is incredibly exciting. In addition to fueling strategy work, it also drives emotional connections with consumers, bonding them to brands in meaningful ways. It is the future of marketing.
Now past the experimentation phase, I can say confidently that AI will be a key driver of technology growth over the next decade and will significantly impact consumer marketing. Initial predictions show the market for AI-driven products and services will jump from $36 billion in 2020 to $127 billion by 2025*. (*Source: BofA Merrill Lynch Global Research Estimates — 2017 the year ahead: artificial Intelligence; The rise of the machines.)
Most AI we work with today is categorized as Artificial Narrow Intelligence (ANI). This means that the AI is extremely adept at executing specific tasks.
Right now, there are seven subsets of artificial intelligence, outlined below. Brand marketers can better uncover insights, connect with consumers, and redefine customer experiences using this innovative technology.
Machine learning (ML)
ML uses human coded computer algorithms based on mathematical models. Probability models then make assumptions and/or predictions about similar data sets.
Currently, machine learning can be leveraged as a service to accelerate sentiment analysis and domain-specific insights. It also serves as a foundational element for identifying consumer behavior based on occasions, perceptions, and attributes to construct themes and trends from unstructured data which represents the thoughts, behaviors, and preferences of consumers taken directly from their online activities.
In 2017 and beyond, I expect more third-party providers will offer ML as a cloud service brands and agencies can leverage to transform products and services into smart objects, able to predict needs and preferences.
Machine learning solutions have allowed my team to align our proprietary structured data assets with unstructured data to combine the best of both worlds. This began to accelerate our processing and analysis capabilities to uncover consumer truths within unstructured data to further fuel our agency’s strategic storytelling.
Cognitive computing builds on machine learning using large data sets.
The goal is to create automated IT systems that can solve problems without human intervention. Marketing centric cognitive computing solutions can consist of a single, all-encompassing solution, or be comprised of multiple services that build and scale applications over time.
From a marketing application perspective, cognitive computing-based solutions range from customer experience enhancing chatbots to closed loop systems for tracking media performance.
Bank of America recently launched the Erica bot using AI, cognitive messaging, and predictive analytics to further influence consumers’ ability to create better money habits.
Cognitive computing will be key to unlocking the potential of conversational experiences. As ecosystems continue to rise, many of the 30,000 chatbots on Facebook Messenger are powered by AI services.
Facebook’s own M virtual assistant housed within Messenger will soon come out of beta testing and will incorporate cognitive suggestions based on content of a conversation users are having. The goal is to make Messenger-based interactions more convenient, enabling users to access services without leaving the conversational thread within Messenger.
Speech recognition and natural language processing (NLP)
NLP refers to intelligent systems able to understand written and spoken language just like humans, along with reasoning and context, eventually producing speech and writing. NLP plays an essential role in the creation of conversational experiences.
Voice-based experiences, such as Alexa’s voice services (AVS), will become pervasive over the next few years. It is projected that by 2020, 30 percent of web browsing sessions will happen without a screen.* (*Source: Gartner analysts at Symposium/ITxpo 2016.)
The core of the AVS experience is a combination of automated speech recognition, natural language processing, and a cloud-based AI that comprise a voice-based user experience.
As with most artificial intelligence entities, learning new skills is how personalized and contextual experiences will be created. With Alexa, it is possible to “teach” new conversational elements and interactions through developing skills.
Here is an example from Domino’s pizza that allows consumers to order pizza directly through Alexa voice services.
Alexa skill development is one of the quickest ways for brands to connect with the rapidly growing audience that calls upon Alexa to empower their daily lives.
Fitbit is another brand leveraging Alexa-based skills to extend brand engagement. Traditionally Fitbit users depend on an app to visualize their data. With the Fitbit Alexa skill users can get a quick update on the stats that matter the most without the need of a screen.
Deep learning builds on machine learning using neural networks. Neural networks are statistical models directly inspired by, and partially modeled on, biological neural networks such as the human brain. The use of neural networks is what differentiates deep learning from cognitive computing.
Deep learning is currently redefining Google’s approach to search, and search engine optimization (SEO) will never be the same. Previously, Google search results were based on algorithms defined by a strict set of rules and SEO was based on regression models that looked at past behavior to adjust a given strategy.
With the introduction of RankBrain, Google’s machine learning technology, in 2016, search algorithms are now enhanced with artificial intelligence. Google is now processing roughly 15 percent of daily queries by mixing the core algorithms based on each search type.
The system is adept at analyzing words and phrases that make up a search query. It also decides what additional words and phrases carry similar meaning.
Expect the percentage of search queries handled by AI to significantly increase. Marketers will need to rethink site architecture, content, and the signals being sent via backlinks as the systems continue to learn on a query-by-query basis.
Predictive application programming interfaces (APIs)
A predictive API uses AI to provide access to predictive models, or expose access to an ability to learn and create new models.
Fortune 500 company USAA is analyzing thousands of factors to match broad patterns of customer behavior through its intelligent virtual assistant Nina.
As we shift from consumers using technology to technology enhancing consumers, predictive APIs will play a key role in providing recommendations, enhancing customer service, and providing real-time analytics without in-house data scientists. This is key to unlocking new forms of value exchanges with consumers in a hyperconnected world.
Image and object recognition
Image recognition finds patterns in visually represented data, pictures, and objects. Facebook and Google are two organizations focused on AI research and solutions in this area.
As image recognition is extended into video and live broadcasts, it will redefine contextual relevance, categorization, and automation of content distribution.
Combined with the advancement of cameras, image recognition and machine learning are transforming the way we process data, including much more than just attitudes and behaviors.
Brand marketers can now leverage images, facial expressions, body gestures, and data collected from IOT-enabled devices to understand the triggers behind behavior and build experiences that anticipate their customer’s needs. This requires brand marketers to transform their data strategy to expand beyond first- and third-party data to also incorporate unstructured datasets that capture affect and unconscious data inputs.
Snap’s pending patent on object recognition is potentially game changing. A recent patent application shows its desire to built object recognition into snaps that can enhance recommended and sponsored filters most likely powered by an AI-based system. This showcases how any object can be aligned with creating immediate context with a consumer and brand.
Olay launched an AI-powered Skin Advisor that ingested user generated photos and provided recommendations for suitable products.
Dynamic experience generation
AI-based systems not only have the ability to parse through large data sets and offer predictive solutions, but also can drive the creation of dynamic experiences. AI will become a powerful tool for creating vs. analysis.
Many startups are leveraing AI APIs to create intelligent solutions. The Grid (https://thegrid.io) is leveraging AI to automate web design with Molly. Molly analyzes design decisions and creates new web experiences.
Eventually, AI will be a key driver of creating augmented reality experiences. Dynamic experience generation through AI will recreate physics, recognizing gestures and movements that can generate new consumer experiences.
Below, Mark Zuckerberg discusses the future of AR/VR at Facebook’s F8 conference.
The various subsets of artificial intelligence will continue to be interconnected, redefining how we approach connecting with consumers. AI makes it possible to know the consumer better than ever before. If approached correctly, with the right mix of AI subsets leveraged, companies will see their business grow.
This is a repost of my recent iMedia cover story.
Follow Tom Edwards @BlackFin360