Marketers Rush to Rise on the AI Tide

Artificial intelligence is the wave of today and tomorrow, and content marketers are struggling not be left in its wake. Among those bridging the gap between marketing and technology is Christopher S. Penn. He co-founded the machine learning and analytics consulting firm Trust Insights along with Katie Robbert, a data scientist and machine learning practitioner. Penn talked about artificial intelligence and content marketing with digital marketing expert Madalyn Sklar, bluntly discussing challenges and what AI can -- and cannot -- do. “Today’s content marketing has four key problems: strategy, tactics, execution and measurement,” Penn said. “Marketers start with execution, then choose tactics, don't measure anything, and then wonder why they have no strategy. “Marketers also make content based on superficial data analysis at best, guesswork, ‘instinct’ or ‘This is what we've always done’ at worst, typically guided by the HIPPO -- highest individually paid person's opinion -- rather than data-driven research,” he said. After that stinging indictment, Penn prescribed orderly questions for content marketers to address:

• What's the strategy, and how does it align with business goals?
• What are your tactics?
• How will you execute the tactics?
• How will you measure?

“By far, the biggest challenge is sticking to a plan,” Penn said. As he sees it, artificial intelligence promises three things: acceleration, accuracy and automation. “Faster, better, cheaper,” Penn said. “Any repetitive task is fair game for AI to tackle and improve. How long does it take to do simple content-marketing tasks like staging, editing or publishing content? How long does it take to research and dig into existing trends, topics and so on? “AI solves many of these problems by doing the grunt work for us,” he said. “Once we train our software and build a model, we can get to work faster and create better content.” Penn agreed that sometimes AI produces better than employees. “Exactly,” he said. “When the machine generates better content than the intern you gave it to, AI is the winning solution. That's content marketing today at too many companies.”

Ripe for automation

Penn gave several examples of processes that can be easily automated using artificial intelligence:

• Research key topics. I'm in the middle of building a new marketing trends report, and I've downloaded 40,000 articles on marketing. AI -- specifically text mining -- is helping mine what the most prominent trends and topics are in that corpus of text.
• What content works. Using a type of machine learning called Markov chains, I track which pieces of content assist conversions on my owned media properties the most. Then I optimize the daylights out of those pieces.
• Network graphs. Using Talkwalker’s data, I graph relationships between entities, then use machine learning to assess the likelihood that a Twitter account is automated. That adjusts influence scores downward based on percent automated.

“The graph is the eye candy,” Penn said. “The value to a client is the spreadsheet they get of exactly who to reach out to.” Leaders should acquaint and train their teams on artificial intelligence. Resistance is futile -- and self-defeating. “Preparing for AI as a leader means cultivating the talent,” Penn said. “Data science and AI skills are rare and in high demand. A good data scientist will set you back $200,000-plus in salary. Build talent internally as much as you can.” He described three key roles entrepreneurs will need:

• Developers to help extract data from systems.
• Data scientists to prep, clean, analyze and model the data.
• Marketing technologists to take the model outputs and put them to work in the business.

“If you don't have the talent, look into training plus tools like IBM Watson Studio,” Penn said. “They provide environments more appropriate for business users. Look for repetitive tasks to automate first.” Artificial intelligence also demands training by marketers. “Marketers need to become the chief questions officers of their organizations, looking for interesting mysteries to solve with data,” Penn said. “That means getting away from ‘do’ a little bit in order to spend time really thinking and being curious. “Marketers must become data-driven,” he said. “Anyone who can't think with ‘both sides of the brain’ is in serious career trouble. Be creative and be analytical. That's the only way you'll grow your career in a world where machines can do pure analytical without you.”

Jack of all trades

Mastering only one skill will not be good enough. “Marketers need to become multidisciplinary,” Penn said. “It's easy to automate tasks within a narrow domain, but you can't blend different domains of expertise together with AI. If you know writing, psychology, music and marketing, you're a lot harder to replace.” Although “artificial intelligence” sounds mystical, it is not all-powerful. There are tasks it cannot perform … for now. “The list gets shorter every day,” Penn said. “Machines can now write credibly well, and that will get substantially better in the next 12 months. Expect machines to crank out decent blog posts, for example, at a massive scale. “Machines can't think across domains,” he said. “That means AI will have a hard time with empathy, with human judgment tasks that ‘break the rules’ and with multi-domain expertise.” People still have roles to play. However, they should not push their luck. “Machines won't replace most human relationships except when your customer experience is so abysmal that dealing with a machine is preferable,” Penn said. “If a chatbot is an upgrade to your service, your service sucks. Department of Motor Vehicles, I'm looking at you.” Content marketers would be well served to remember that “machines cannot do general life experience.” “General life experience is broad and contains lots of different inputs,” Penn said. “Machines can't replicate that yet. It requires real sentience that they don't have. What makes us human is a mix of weird data mashed together with pattern recognition. “Until machines can do massive pattern recognition across many different, disparate inputs, tasks that require that kind of knowledge -- like empathy -- are relatively safe,” he said. That might be only a short-lived saving grace. “There will come a time when we have to contend with a machine being able to do empathy,” Penn said. “When that day comes, it will be far more than a marketing problem. But we'll deal with that when it happens.” For more about Penn’s work, he has published “AI for Marketers: An Introduction and Primer