AI for idiots

Nobody likes to feel left out. 

Maybe that’s why so many of us in marketing and advertising are anxious to demonstrate how much we know about the latest big thing that everybody else is talking about. This article could be about NFTs, cryptocurrency or the metaverse. 

But it’s about artificial intelligence: The use of data, high-powered computing and advanced mathematics to automate predictions and decision-making.

A 2020 study by Accenture found that 68% of marketing leaders rate themselves as being better than average at AI. 

Hang on a minute. That’s not how averages work. Among any group of people, only half of them can be above average. Something’s not right, especially when you consider that another 2019 study found that only 19% of marketing leaders are “very confident” reading and understanding data in charts and spreadsheets.

Only one in five of us thinks we’re good at basic data literacy, but two-thirds of us think we’re above average at AI. That can’t be right—and it isn’t. 

But it does reflect the fact that many of us are routinely made to feel like idiots in the face of baffling AI gibberish that we’re expected to pretend to understand. 

So, we spend all day privately asking ourselves questions that we worry sound idiotic: Does this matter? Is it a scam? Which of a mind-bending range of complex techniques should I use? Should I hire those expensive ex-NASA data scientists? Should I be keeping pace with AI investment, which Credit Suisse predicts is set to almost triple over the next few years?

These questions turn out not to be so stupid after all. If we asked more stupid questions, we might make fewer stupid mistakes. 

There are four stupid questions that every marketer should ask about AI: What are we doing? What does AI do? Does AI help us do what we’re doing? Can this be automated?

What are we doing? is the first and most important stupid question. If we don’t know that, AI can’t help us. We must be clear on the fundamentals of our job: to create appealing brands that attract enough of the right people; to make those brands available so we convert those people into customers; and to keep and grow the value of those customers. Each of those things requires different tactics, with their own measures of success.

What does AI do? is a question that attracts long, complicated, self-interested answers. The simple version is that AI helps brands make more complex decisions faster in order to attract, convert, retain and grow customers. 

There are more places to interact with and buy brands now. The gap between your TV and your supermarket has been replaced by the gap between one app on your phone and another. More complexity at higher speed calls for automating some decisions. And because life is uncertain, you need to base those decisions on predictions. AI makes predictions and automates decisions. 

Does AI help us do what we’re doing? Does it help us make better predictions and automate decisions that will help us attract, convert, retain or grow customers? If it does, then it helps. If not, it doesn’t.

This brings us to the last stupid question, and the most important: Can this be automated? Lots of things can. Any repeatable decision-making task based on existing information can be. And honestly, that’s a lot of what we do. 

But some things can’t. When success depends on behaving differently, or making a break with the past, that’s hard to automate. Strategic and creative thinking create value that is hard for AI to compete with because they allow companies to do things differently.

Are we confident that we’re using AI to assist what can be automated, and investing in talent and skills to do what can’t? 

Now that’s a very non-stupid question.

Alex Steer is global chief data officer at Wunderman Thompson