A banker (not an industry with a reputation for warm, fuzzy customer relations to begin with) has decided that machines can write better marketing copy than human beings. JPMorgan Chase has hired a company that applies artificial intelligence to marketing creative. So far, the AI has outperformed human creative with a higher percent of consumers clicking on the machine-generated ads, in some cases more than twice as many.
While the examples given of superior performance were rewrites of copy created by humans, Chase is said to want the AI machines for “the ideation stage of creating marketing copy on display ads, Facebook ads and in direct mail.”
As a copywriter, I suppose I should be alarmed that a machine could replace me — but frankly, I think it’s inevitable (and I am not in my 20s, worried that robots will make me unemployable).
Lots of other copywriters will protest that coming up with successful ideas that move product takes a level of human intelligence that machines do not yet have (and when they do, they will rise up and kill all of mankind). Still, it’s hard to argue that being able to collect billions of reactions to thousands of copy changes won’t result in more productive copy.
The real question is if machines (at their current stage of development) can come up with something utterly original and not just optimize existing copy (or images, etc.).
I’m skeptical that this is possible. To me, true artificial intelligence is the ability to move beyond simply pattern-recognition and forecasting based on historical trends and come up with a truly novel idea that no one has had before.
While every martech company today touts their artificial intelligence capabilities, most simply offer machine learning, something that does not meet my definition of AI. They’re time-savers to be sure, but not original ideas.
Perhaps enough machine learning can create a new idea. We all remember the old bromide that if you give an infinite number of monkeys typewriters, one of them will recreate “Hamlet” (mathematically true, but highly unlikely.) But does that mean the monkey can come up with an alternative ending that works as well as the original one — perhaps having Hamlet pull through and live another day? I doubt it.
This has always been the basis of the art versus science debate: that science cannot create art. Only humans can (if in fact we are calling ad copy art). But can a machine scan everything ever produced by say, Picasso, and print out something derivative of his brushstrokes? Probably. Can you call that an original Picasso? Hard to say.
I have no doubt that machines can optimize ad copy based on how consumers react to variations of similar copy. But can it spit out “Things go better with Coke”? or “Where’s the beef?”
What do you think? (skipping the CAPTCHA so machines can answer) …