Fantasy and FOMO: Why Marketers Need to Question AI
Picture yourself as a prospector in 1859 drilling one of the first commercial oil wells in western Pennsylvania. You seek a dark, mysterious substance that can illuminate rooms at night. Your peers are drilling too, so you assume there’s value. To be an ‘early adopter’ is both thrilling and comforting. You pursue a fantasy of success fired by the most ancient of fuels: FOMO (fear of missing out). You haven’t a clue how oil will affect our environment and geopolitics in 100 years, but at your moment in history, it’s too enticing to resist.
That is the state of artificial intelligence (AI) in the marketing industry. In Widen’s 2018 Connectivity Report, we found that marketers are seriously confused about AI, yet have a giant appetite for it. As a community, we must question the fantasy and FOMO driving this trend.
Voracious and Confused
In early 2018, my team and I conducted 32 phone interviews and 506 online surveys with marketing professionals. 86% of respondents told us that they’re not using artificial intelligence in marketing and creative work yet. For comparison, Gallup finds that almost 85% of consumers already use AI in their personal lives, whether they realize it or not. Seemingly, marketers are behind.
The gap is surprising because the top trend in our study was “personalizing the customer experience.” 93% of respondents feel that personalization at scale is attainable, but 58% are unsure of how to achieve it. Personalization at scale is probably a pipedream without technology that segments customers automatically. We might call that technology AI (although as you’ll soon see, we should be careful with that label).
When we asked these same participants what “artificial intelligence” means, over 50% said it either reminded them of futuristic movies and robots or that they didn’t know. Hence, the oil metaphor. Analysts, vendors, and reporters told us that AI is the ‘next big thing,’ and we bought in. They sold a fantasy about the wonders AI will accomplish, igniting the FOMO furnace. What are we really buying?
Good news for marketers: everyone is confused about AI. Experts don’t even have good definitions for the word “intelligence” let alone “artificial intelligence.” And bad news for marketers: if few people understand what AI is, we’re falling for own tricks. AI is a locked-and-loaded buzzword.
The definitions are problematic because most assume that human beings are “intelligent,” a word our species invented. In our limited experience of the universe, we don’t know of any organism that exhibits greater “intelligence” than us. Our biology is the reference point for machine intelligence.
So, the experts have been debating definitions for decades. One article by researchers at Deloitte cites three different definitions just to make a point. Even the definitions that sound ‘right’ don’t make total sense. For example, Deloitte says, “AI is concerned with getting computers to do tasks that would normally require human intelligence.”
That’s reasonable if we’re considering tasks like voice recognition, driving, or playing chess. Often though, we’re trying to get AI to do things that human beings cannot do. For instance, the recommendation algorithms behind Netflix, Amazon, Facebook, etc. find correlations that human data scientists would never notice. Eventually, those algorithms might predict, with amazing accuracy, what we will buy next. Human beings aren’t known for accurately reading the future.
The Burden is On Us
The struggle to understand AI means that marketers—and many other professionals—could make far-reaching decisions about concepts they don’t understand. Vendors may still be working on their definition of the technology.
Thus, we have a responsibility to question AI technologies served to us on the silver platter of buzzwords. I can’t supply a better definition than the experts, but I can arm you with five questions for vendors:
- What do you really mean by AI? What makes your tool(s) intelligent?
- Can you explain the difference between the AI you claim to have in your system and general machine learning, which trains AI (i.e., by giving algorithms instances of a pattern, like red traffic lights, so it learns to identify instances of red lights on its own)?
- If the AI behaves unexpectedly and does something to embarrass our brand, what will you do about it?
- How does your AI use the data our company collects? Why should our customers trust that their data is safe in your system?
- In five years, what do you hope your AI (if that’s the right word) will be able to do? Where’s the roadmap aimed?
We may develop a dependence on a resource that, like oil, changes the global economy but not without unintended consequences. These questions probe for consequences, or at least gauge whether the seller understands the nature of the product. You want vendors who are mindful about AI and how it can fail.
Marketers are among the frontier prospectors drilling wells to strike it rich with personalization and other innovations in customer experience. Whatever the merits of our collective fantasy, FOMO makes it hard to pause and scrutinize our actions. Let’s do the hard thing. Stop and question “AI” before you implement it.
Do you agree with the questions I’ve outlined for vetting AI before diving deeper into implementation? I’d love to hear your thoughts in the comments.