Take my job! A case for automating creativity

Automation won’t just make us faster and more efficient, but also more intelligent and creative

The robots are coming. Indeed, they’ve already come, settled, and spawned their successors. Now, do we rise up against them as conquerors or welcome them as liberators?

Jobs that were once the bread and butter of American workers are being automated at an increasingly accelerating pace. This has already led to substantial job redistribution, and there’s no shortage of voices prophesizing future unemployment. Cynics estimate job losses at 5 million by 2020. The more frantic among them predict unemployment as high as 75% by the end of the century. That’s likely overstated, if not wrong altogether. 2020 is fast approaching, and unemployment is steady at an almost 17-year low.

Not my job

You’re probably thinking, “I’m safe anyway. A computer will never be able to do my job.” Well sure it can — at least aspects of it. So far, the discourse around automation has focused on industrial, clerical, and service jobs. And many argue that automation will never fully enter the creative industry, but I guess that really depends on how you define “creative.”

Does it take creativity to write a news report? Does it take a refined ear to identify a future hit song? Does it take a spark of transient ingenuity to compose a symphony? If so, then creativity can be — and already has been — automated.

Nearly three years ago, the AP implemented an artificial intelligence program that writes earnings reports. Within the first year, it published more reports than its human overlord had her entire career. The New York Times also used an AI program to help generate a feature story around the same time. It composed 52,000 unique narratives that deployed automatically based on location.

Many industrial applications are still in their infancy, but that’s changing rapidly too. Today, startup Persado is using AI to write action-optimized copy for businesses. And that’s not just optimized in aggregate. It promises to target users all the way down to the individual with “personalized language and emotion.” I’d be skeptical, too, if Goldman-Sachs hadn’t led a $30 million investment round in 2016.

Disrupting or eliminating?

It’s natural to find this trend worrisome or flat-out terrifying. You wouldn’t be alone. While plenty say we’re approaching doomsday, those apocalypse-hawkers have made the same tired argument for decades. There’s an equally compelling case that automation will do more than just push down prices and displace some tedious work.

Automation has consistently grown productivity and created new jobs that are more interesting, more creative, and less harmful. Does anybody want to laboriously create spreadsheets by hand? How about process payments? Just as automation and information technology have streamlined many of the monotonous administrative tasks we used to do, so too will it eliminate many monotonous — though more complex — tasks we are currently doing.

It also has the potential to grow our industries and enhance our individual abilities. Persado, for instance, isn’t eliminating human jobs. It’s growing the industry by creating a new category of marketing that wasn’t possible before. The New York Times isn’t firing its writers in favor of automation. Instead, it’s allowing its current writers to tell deeper, more complex stories.

Apply liberally to healthcare area

The applications are nearly limitless for any industry, but healthcare marketing lends itself particularly well to this emerging technology. That’s because many of the communications we develop and publish are fundamentally based on information parsing and structuring. That’s where automation shines.

It’s not unreasonable to imagine a program that could parse and pull claims from a source document. Scientific papers follow a regular format, and they predictably structure information in tables and graphs. A full claims document could be developed at near instantaneous speed, with objective evidence stored in the metadata to support its validity.

Taking it a step further, the program could iterate each claim with various emotional signifiers. Then, it could automatically A/B test them. The output may not be a full-fledged message architecture, but it would be an incredibly useful resource for a strategically sound, evidence-based message architecture.

Getting a little more fanciful, you could probably imagine a program that creates and tests concepts. By analyzing product categories, as well as the media that a target demographic consumes, a program could identify and pull potential imagery. With an A/B test, the most relevant and impactful images could be identified. It wouldn’t result in a wholesale concept, but it would be an amazing starting point.

Automate it! Today?

Where is this revolutionary AI and automation? It feels perpetually 5 years away. But it’s not. These technologies are in use already, albeit at a low bandwidth. Any copywriter worth her salt has a favorite thesaurus and idiom generator. That’s a kind of automation. I’m not good at coming up with every idiom that relates to salt; let a computer do it.

Agencies as institutions could deploy part of the Shining Agency of Tomorrow, today. Take the process used to route documents. It’s digital, sure, but it hasn’t fundamentally changed from its paper origins. As a result, the Editorial process remains as taxing and time-consuming as ever.

For the very narrow-use case of a “check change,” a program would be objectively better at finding, marking, and routing errors. A pixel-by-pixel comparison between two documents would take seconds. A program that simplistic wouldn’t require more than a couple dozen lines of code. A more complex one could distinguish errors from changes and route them only to the appropriate stakeholders.

But that’s not an endorsement for the elimination of any job or department. The human element is invaluable. Even in this narrow-use case, the program wouldn’t cut editors out of the process. It would only offload tasks that computers are better at, allowing our editors to focus less on pixels and more on global elements, like consistency and style.

The same would hold true in every department. None of these examples would eliminate any one position. However, they would offload many of the technical or monotonous tasks that we have to do in order to do the things that we like to do. And that would free up time and energy that we could use to think more deeply, strategically, and creatively.

At some point, the creativity may be in how we set the parameters and edit the results. Maybe we’d choose messages we never would have thought of, or maybe we’d just get there more quickly. And maybe we’ll invent new things to say, and bigger stories to tell altogether, making healthcare a richer, subtler space. So, robots, take my job! At least, take parts of it. I’m not so good at them anyway.