Don’t Replicate Human Intelligence, Augment It

Why must we always make the perfect the enemy of the good? The drive for Artificial Intelligence to solve a world of problems for communicators will likely result in failure—at least for the foreseeable future.

Computers can make our lives easier, but they won’t relegate PR pros to a life of sitting in a La-Z-Boy anytime soon.

But that doesn’t mean that technology won’t substantially change how we do our jobs.

I prefer to think of AI as “augmented intelligence” because I believe that we can use the power of silicon chips to increase productivity and produce better results when paired with good old-fashioned human brainpower.

It turns out that I’m not the only one who thinks this way. Alas, I don’t get to take credit for inventing the term “augmented intelligence.”

In fact, one of the recognized leaders in the AI field—IBM—says that they, too, prefer “augmented” to “artificial.” In response to an Obama Administration request for perspectives on the future of AI, IBM argued that the word change recognizes the “critical difference between systems that enhance and scale human expertise rather than those that attempt to replicate all of human intelligence.”

In the world of media and communications, we have seen many attempts to go the “traditional” AI route of replacing humans. Whether that’s a system designed to substitute a computer algorithm for human judgment in gauging the sentiment of a piece of content or having a program turn raw data into news articles about sporting events or corporate financial results, it ultimately results in something far less accurate and satisfying than a human-generated solution.

Take those much-hyped programs that convert a baseball box score into a game story. Does it look authentic at first blush? Absolutely. It follows the typical formula that newspapers have used for the last century or so to encapsulate the results of a game.

Yet it lacks the color and perspective that an actual journalist might lend to that story. Nowhere in those pieces can it capture the fact that the starting pitcher left due to an injury or the light-hitting second baseman rocketed a homer into the upper deck.

Imagine how that same story might be more appealing if the computer “wrote” the first draft and a human added that color. The reporter could then be more efficient, but the final product would still have that additional touch that would help the reader better understand the game.

This isn’t really a new concept. Ronald Reagan (and others) famously took ticker tape reportsfrom baseball games that they were not able to attend and turned them in to colorful radio broadcasts. That was “augmented intelligence” of a sort—early 20th century-style.

A similar augmented intelligence approach can help when it comes to evaluate communications performance.

While technology doesn’t yet exist that can adequately replace human coders, there is technology that can help make those coders more efficient. That might include allowing automated sentiment to take a first pass at an article, leveraging entity extraction to properly classify stories, utilizing algorithms to identify anomalies, and employing technology to enforce business rules for consistency and accuracy.

These all represent important advances that will help to make proper measurement more accessible and reliable.

Technology has made—and will continue to make—all of us in the PR and marketing fields increasingly effective and efficient. The pace of developments continues to accelerate and will open the doors to ever more innovative tools and solutions. They will help us avoid errors, shed routine tasks, and develop deeper insights.

But if we focus too much energy on replicating human intelligence, we will move away from the practical solutions that IBM itself argues make more sense. If the House of Watson shies away from using the term “artificial intelligence,” perhaps those of us who don’t have the same resources and track record should take note.

A version of this article originally was published by The Measurement Standard.