There’s a certain face people make in cars that wasn’t seen on US roadways 10 years ago. It’s a combination of confusion, anger, and amazement. If I had to give it a name, I’d call it “Misunderstood Technology Face.”
You know it. It’s probably happened to you when you type in an address on Google Maps or Waze and the route travels down side roads, back alleys, into drainage ditches and through suburban backyards. You can’t imagine how that’s going to get you where you’re going. You don’t trust it, you go the way you’d normally go, and it takes you twice as long. You should’ve trusted the app.
Technological advances come along so fast now, especially in the digital space, that some mistrust is expected. We especially don’t trust machine intelligence (MI), in part because we don’t understand it, but also because of all the fear-mongering about doomsday, regardless of how accurate those theories are. That’s the trend with MI, and with new technology generally, probably from the time someone first rolled a wheel down a hill: we hate it until we get it, and then we can’t imagine life without it.
When MI makes its way into the business world, we see the professional equivalent of Misunderstood Technology Face. Those who figure it out, though, have a distinct market advantage over those who don’t.
The “Why” Matters
The goal is visibility. Machine intelligence will bring together data from all corners of a company, making what was previously siloed and hidden available for deep insights into how the business is (or isn’t) working. But here again, these insights are only effective if the user understands what the heck the MI is actually doing.
If you’re in sales, the future may well lay in predictive lead scoring. To determine the best prospects, an MI-powered system takes all the available data — patterns in previous sales interactions, keywords in emails — and extrapolates a list of who’s most likely to buy in the near future. These leads are scored and presented to the sales rep. But that rep has to be able to parse that information, to see what elements led to those conclusions. How can you trust a list of names if you don’t know how it was put together? Transparency, in other words, is paramount.
When the user understands how an algorithm arrives at an answer, that user can apply it most effectively. Say your list of prospects is the combination of 1) how recently the prospect called plus 2) the number of emails exchanged weighed against 3) how many other previous prospects have bought in these circumstances. That knowledge allows you to reach out with a fuller grasp of the prospect’s circumstances. Your (very human) sales instincts can be brought to bear in the best possible way.
The Human Element
Some fear that human input will be completely overwritten by algorithms. The truth of the matter is that human intuition will complement, and in fact complete, the cycle. Intuition is an essential component.
Across the board, we’ll see MI affecting how the entire business works. In service, a bot could take care of rudimentary questions so that service agents are freed up to handle more complex issues. In marketing, MI can simulate A/B testing, saving time and testing resources. And in sales, data yield best-case scenarios.
In all these cases, though, human intuition (and training) sits at the top. It’s that final, essential element that finds customers, wins them over and helps them solve problems down the line. That insight, then, is fed back into the system to create smarter algorithms and better systems. It’s a feedback loop designed to serve employees of the company and their customers.
Learning from Intuition
The human instinct is so important to the field of machine intelligence it’s spawned an offshoot called “artificial intuition.” In one experiment, MIT students played with different ways to optimize airline routing. The smartest solutions were better than anything the algorithm could come up with. So the researchers incorporated those strategies into future algorithms.
Whether it’s a map app, a sales algorithm or airline routing, the success of the system relies on cooperation. People must not only use that system; they also have to understand its workings well enough to make decisions informed by the system’s algorithms but not necessarily limited by them. And then, crucially, the system has to be designed to learn from whatever decision the person made, and its ultimate outcome.
There’s a significant amount of trust and reciprocity in making machine intelligence work as well as it possibly can. In example after example, the machines augment human performance rather than restricting or replacing it. It’s what at least one company calls “co-bots.” The most successful companies will leverage advanced technology and human intuition in equal measure, built on a platform of trust.
