Both Ends Of The Data+AI Spectrum

In late March, when the so-called “scandal” broke about how Cambridge Analytica used Facebook data, there was a story in The Wall Street Journal about how crappy most companies’ data is for actually understanding their customers.

To me, these two stories represented two ends of the same data+AI spectrum, and together say something important about the potential impact of artificial intelligence and machine learning algorithms on the competitive landscape of just about every business.

(First, though, forgive me for being cynical about the current hue and cry over what Facebook knew when and how its terms may have been violated, since the details of what Cambridge Analytica did have been widely circulated and boasted about since the 2016 election. I wrote about it myself in February 2017. It’s hard to justify as a scandal when the participants are out there every day spreading the word in public like Jehovah’s Witnesses.)  

But I digress.

Data+AI presents the classic garbage-in/garbage-out challenge. That’s why you may have read about “racist” AI, for example. You train an AI using crappy data, and you get crappy results. So there are reports of facial recognition algorithms not recognizing black faces (they were trained on mostly whites) and a criminal risk-assessment system used by U.S. courts flagging blacks almost twice as often as whites (revealing the inherent human bias in the data on which it was trained).

Businesses don’t have to be racist to have crappy data. In fact, because businesses are populated and managed by us flawed humans, most of the data at their disposal is, well … crappy.

The Journal story noted that most organizations’ customer data is entered by salespeople. And as Derek Stewart, chief strategy officer of Stein IAS, told me: “Salespeople are never going to be objective. They’re only ever going to tell you what they understand about their customer that helps them hit their targets. They have a very myopic view on a certain thing. We’ve got to understand that bigger picture.”

Stewart strongly believes that ultra-deep customer insight — what he calls the audacious goal of knowing customers better than they know themselves — is the key to success in what the Stein IAS folks call the post-modern marketing era, arriving as we speak. That’s going to require very high quality data and machine learning algorithms to leverage it at scale.

Up to now, though, prosperous businesses owed their success to the still-unique ability of the human brain to assimilate flawed data and intuitively filter and analyze it to produce a reasonable outcome. Artificial intelligence can’t do intuition yet, and likely won’t for a long time.

Cambridge Analytica, however, demonstrated the awesome power of really good data+AI.

Today, the number of companies that could pull off excellent data+AI is probably a rounding error in the U.S. economy. But the Facebook-Cambridge Analytica story should be a big wake-up call.

Legal or not, what Cambridge Analytica did demonstrates what’s possible, and the Journal story shows how bad most companies are at this. Now, companies should begin to “get it” and start focusing on the excellent data+AI end of the spectrum (while either staying on the right side of privacy law or covering their traces!).

And if others fall too far behind, will they ever be able to catch up?

The good news is, there's still time to act. Nick Jordan, CEO of — which you can think about as the Airbnb for data in the AI age — told me that “It's still so early in AI that I don't think the winners and losers are being decided now. Friendster was a thing before MySpace was a thing before Facebook was a thing. I think we're closer to the Friendster of AI than we are to the Facebook of AI.”

As we wrote before, Jordan’s start-up hopes to make it easy for companies to access and integrate multiple high-quality datasets from other organizations.

The moral of this story is to start cleaning up your data act now, regardless of how soon you plan to consider AI. It’s the smartest thing you can do today to prepare for tomorrow.

Next story loading loading..