Does Marketing AI Replace People or Enhance Them?
28 November 2017 | 10:03 am
Blame Hollywood. Blame Elon Musk. Blame whomever you want for the wide perception that Artificial Intelligence will put us all out of work. Others say that AI will lead to a life of leisure. Few are pointing out that those two predictions are the same–it’s just a question if you are optimistic or pessimistic.
But it is overly simplistic, because the all-knowing AI presence–Artificial General Intelligence–is so far a figment of imagination. Today, we are benefiting from Narrow AI–machine’s ability to outdo humans at just one thing, such as chess, or Jeopardy or Go. These AI wonders would be left wondering if applied to any other task.
And most AI in marketing is not even as autonomous as the game-playing types that make the news. By far the most prevalent AI in use is “human-in-the-loop” AI, such as semi-supervised machine learning. Rather than the computer doing it on its own, it is human beings that help shape the computer’s judgement. I work with Converseon, an AI-based social listening company, which uses human-coded data to do its initial training for sentiment analysis. But as it makes predictions, it uses its confidence level to decide which calls it is sure of and which ones it will refer to human beings to check. Any corrections are rolled back into the training data to make it even smarter.
That approach is more likely to be how AI is used today. Rather than eliminating people, it needs people to train it and people to correct it. It can outperform people over time, but its initial usage is to augment the performance of people. If you’ve been waiting for AI to wipe out your marketing team, you likely have a long wait. But if you want to use AI to make better decisions, the future is now.