IBM’s Michael Trapani: AI may never understand a great meal, but it understands performance
Previewing his MarTech Conference presentation, Trapani sees AI not as a team leader, but as a team specialist.
“How AI will let us be marketers again.”
That’s the topic that IBM Global Product Marketing Leader for IBM Watson Michael Trapani will tackle at our MarTech Conference in Boston in early October.
Like a good sidekick, artificial intelligence (AI) now and in the future can tackle all the dreary — and impossibly complex — tasks so that marketers can get back to the fun stuff, like creating campaign stories or orchestrating a groundswell of demand for a new product.
But, I recently asked Trapani, where are the limits for AI in marketing?
Once IBM Watson and his many artificial colleagues have mastered the ability to generate insights from data, manage the day-to-day modifications in a targeted ad campaign and orchestrate the personalized messages best designed for each consumer … what then?
Already, marketing automation platforms tout their movement toward becoming self-running. If cars can one day autonomously drive passengers to their declared destinations, navigating all the random occurrences that real-world streets toss up, then why can’t marketing tools completely orchestrate their efforts, once goals have been set?
Certainly, Trapani told me, AI can derive operational rules from historical data — such as for personalization — for almost any activity, up to a stated level of confidence.
AI, he said, is “a member of your marketing team,” where the marketer is the team leader and AI is one of your specialists.
What is off limits?
Just as a human team member’s recommendations must be tested in the field, he said, so must any recommendations from the AI-powered team member’s.
Even if AI ventures into fields that humans tend to think they own — like creativity — Trapani’s position is that such adventurousness is fine, because it will be tested before wide implementation.
Already, there are indications that AI wants that creative role, as evidenced by marketing emails written by Persado’s platform or IBM Watson’s editing of film trailers. If an AI-powered platform comes up with a new subject line and body text for an email campaign, Trapani believes that would be OK — because it would be extensively tested, just as it would if a human team member wrote it.
Similarly, IBM’s corporate position argues that AI can promise “a new level of collaboration between man and machine, and will only augment and expand human intelligence, not replace it.”
But, if AI can augment/expand some kinds of human intelligence, and not replace others, what are those others?
If creativity per se isn’t off limits, what is?
Here’s a possible clue: One of IBM Watson’s many careers has been to act as Chef Watson, where he generates a unique recipe based on ingredients input by a consumer, based on his understanding of human tastes. Ok, it’s a kind of book learning of average human tastes, but, apparently, it is reasonably accurate.
If he’s doing that, exactly what is reserved for human intelligence?
I presented the idea to Trapani that great marketing is about conveying a great human experience — the sound and smell of the steak’s sizzle, the pleasure of seeing your daughter over an iPhone video chat, the kinetic thrill of a powerful new car threading those curves on a country road.
Whatever we can understand about Chef Watson, we know for sure there is at least one thing he has never done, and never will — at least until the possible occasion when he achieves consciousness.
He has never enjoyed a great plate of pasta primavera.
With competition just a click away, the central brand difference that marketers increasingly must convey is the difference in human experience provided by the product.
Trapani recognizes the limitation that AI doesn’t enjoy pasta, but points out that, for marketing purposes, it does have knowledge about the thing marketers truly care about.
“Machines don’t feel,” he told me, “but they understand performance.”
So, the pending question is whether AI’s “book learning” of the phrases, campaign frequency and personalization needed to get the best responses is what a new restaurant really needs in its ad campaign. Or, perhaps a marketer has to actually experience the restaurant’s pasta to convey the difference.