Creative Automation: Software is coming for creative jobs.
There’s a narrative rolling around the tech world and some political circles about the future of work — that many jobs will be displaced in the next ten to fifteen years. The story usually starts with driverless technology and the trucking industry or perhaps digital assistants and AI’s coming servitude. Another outcome of the coming technical revolution hasn’t been shared as frequently and that is that many jobs that will be replaced by software are white collar jobs. And what I see no one saying is this: many creative jobs are on the cutting block.
I see two sides to this transition. As a creative consultant to often cash-strapped organizations and non-profits who find it hard to attract high quality creative talent, the rise of software solutions to meet the production demands of digital organizing teams is a blessing. As a former designer and friend to many creative teams in advertising, entertainment, and politics, I worry about the future employment of lots of people I know and many trying to get into the field right now.
I’ve spent my creative career in entertainment consumer products, politics, and advertising so this is written with a view into those industries. The coming automations, though, will impact all digitally-based creative roles.
This administration has certainly shaken creative talent out of their safety nets of corporate or agency life and given them the fire they need to apply for jobs in politics. The designers looking to work for political organizations right now are the best pool of talent I’ve ever seen. But the few who make it through the interview processes, who commit to the lower pay typical in political or advocacy work, and who become the next wave of leaders in political design will be just that…a few.
“Competitive compensation is critical to winning today’s tug-of-war for skilled creative talent. It’s also a company’s best line of defense against losing top performers.” — The Creative Group
Political and advocacy groups are finding it incredibly hard to attract top creative talent. Many cannot pay to compete with technology companies or corporations or even the consulting firms that serve them. It used to be that the best creative minds went to NYC to work in advertising or LA to work in Hollywood. Now, they go to SF to work in tech…and some of them are literally designing their future peers out of jobs.
But not every great designer wants to figure out how to make the most frictionless button. Some want to use that thinking and apply it to voter mobilization or campaign fundraising. Unfortunately, these designers are few and far between.
The majority of organizations and campaigns will be staffing up with production-focused designers who have little to no experience and potentially no formal training. This hiring practice is common in politics (and the reason politics isn’t typically on the radar of a super-talented recent graduate). Campaigns and advocacy groups need people with basic technical creative skills to execute designs or videos as part of organizing or fundraising efforts.
The 2016 US presidential election and the resulting reviews of content, digital advertising, and Cambridge Analytica gave us a preview into the future of creative automation and it’s impact on politics. Cambridge Analytica is stated to have run as many as 50,000 ad variations a day and upwards of 175,000 on the day of the third presidential debate. Designers cannot produce that many variations without software to assist. Ad traffickers cannot manage that many A/B tests without software. Cambridge Analytica fell apart in the years since because of how they got their data. Exactly how they executed those campaigns still hasn’t fully come to light and that is reason enough to believe that the software will be used again and has likely evolved. (This is assuming that they used software and not a content farm, which might be too generous an assumption.)
Similarly, marketing agencies and creative shops, especially those working in digital ad production or content marketing have to staff up with design, video, and writing talent to produce the work they’ve committed to do for their clients. These are the jobs that will first be replaced.
Design as a career continues to have very low unemployment rates. 2018 is a designer’s market. And while it sounds like there is an opportunity to build a career, I think that this will be short-lived. The gap between talent availability and the needs of organizations to execute digital content production is the gap that technology will begin to fill first. Design and content production will soon be handled by software.
Resizing digital ads is a skill of the past. Most digital ad production can be done utilizing software that doesn’t require an Adobe license or a design degree. Video iterations can be tailored and executed via software that will then serve the right video to the right audience. Agencies that continue to staff for these production roles are missing the writing on the wall and their bottom lines will pay for it in the years to come as production needs continue to grow to satiate the addicted users on the other end of an ever-crowded digital environment.
As the agency model continues to evolve, some big brands are slashing digital advertising budgets or setting up their own in-house teams so they can keep closer tabs on the creative and marketing processes. The ones who invest in the right ad tech now will save tremendous amounts in agency fees and talent costs in the next decade all while continuing to produce the marketing assets needed today.
The explanation of the five autonomous driving levels is a perfect analogy for the creative automation that is coming.
Level 1 — (Creative Assistance) Software can assist but the employee is making the creative choices.
This type of software assistance has been growing for the past decade, lead by Adobe. To watch Photoshop develop from Photoshop 9 at my first ad agency in 2004 to Photoshop CC 2018 in my current iteration has turned me into one of those people who say “back in my day, I had to…” From rulers and alignments, to predictive background fills, software continues to make production faster and easier.
Canva is probably a level 1.5 product that allows users to drag and drop elements of their own or play within existing templates to create digital products. More often used by non-designers, they’ve made a lot of progress in the last 6 years, in terms of quality and professionalism of the content that can be produced using their software.
InvisionApp is another web-based application designers use — often UX and UI designers — that aids in the design production process by providing a wysisyg interface for designers to lean on to quickly prototype their application or web product workflow.
Level 2 — (Partial Automation) Software assists with production but the decisions are still required by the creative employee.
A great example of partial automation happening in creative production is social media content tools like Hootsuite, CoSchedule, or Tweetdeck. These allow the content manager to plan the full month of content, set it all up, and let it run. Some even provide shortcuts to templatize the content creation process, regenerating or reposting previously written content.
Adobe’s Spark can produce professional looking digital content with just a few key inputs from a team — a logo, a tagline. It lets users pick from a few pre-designed looks and it begins to populate social ads, videos, banners — anything, without needing more design software.
IBM and Shutterstock recently announced a partnership to offer a similar solution. Their version connects users to Shutterstock’s deep trove of imagery and music scores to help teams create full professional quality campaigns in their content hub. Just like Spark, they’ve got all of the practical elements taken care of.
Level 3 — (Conditional Automation) Conditional production is the beginning of software-aided hyper-targeted creative.
Spongecell allows agencies to run dynamic ads that can adjust by data points in the users’ browser. Upload a few different background images and a matrix of copy and button language, and their software can match the right message, image, and button to a user in Florida sunshine and send a completely different ad to someone in Colorado’s latest blizzard.
Google has been running conditional banner units for politics for at least two cycles. For example, a unit will say “7 days until election day” on a certain day and then switch programmatically to “6 days until election day” the next day. This is conditional in the sense that it uses variables to determine when to combine different pre-designed elements, saving production and trafficking time.
Flexitive is software that can take a background image or video, text, and a button, and then resize that layout across any number of sizes and deliverable specs. They’re not the only player (Bannerwise, Makethunder, etc.) in the space but they are growing quickly and will soon be the type of software that takes the jobs of anyone who is manually resizing banner ads in Photoshop.
Level 4 — (High Automation) Software is driving the decision making process and must alert the user of problems to allow the user to troubleshoot or take over.
This is where we are today. The combination of programmatic ad buying, machine learning applied to reporting, and software-driven content production (with a social listening component and bot-driven writing) means that a brand or company could train software to develop, run, and re-tool campaigns to continually optimize creative and refresh it automatically when a predetermined threshold has been met or a limited return has been reached.
The creative team member’s role will be to establish loose visual brand parameters, perhaps help with the brand voice, or doubling back on social listening results to be sure that the appropriate input for content is making it into the loop. There is no more designing and pixel pushing. The stories and content being shared will be determined by data — matching the right narrative to the user based on the data points we have on that user and then targeting lookalikes and tweaking the narratives, language, imagery, audio, impressions, channels, etc. from the last data point until we’ve reached every user in the required universe with the exact number of units needed to persuade each and every user to take an action.
Adobe is a company I would keep a close eye on for this kind of software. With their creative and marketing products, plus they’ve picked up commerce and voice companies in the last year, they’re sitting on a pile of tech that has every bit of potential to automate much, if not all, of the digital marketing toolkit. All the usual suspects are playing in this space too- Facebook, Google, Microsoft, Salesforce, etc.
Others could, of course, be working on this but I would imagine they will leverage it like Cambridge did, in a consultancy capacity rather than standing up self-serve software. Either way, creative employees have limited opportunities inside of marketing teams that use software or consultants capable of this highly-automated digital marketing.
Level 5 — (Complete Automation) In creative production, this would be when an input such as a revenue goal or awareness initiative is plugged into software that can then source and produce all the necessary assets for a compelling digital campaign, run the ads, and report back and set up a new flight based on performance metrics from the first flight.
We’re not here, yet. But there’s no doubt that this what IBM, Adobe, Microsoft, and others are working toward. They’d be crazy not to.
This study by Oxford and Yale polled AI experts and they landed on 11 years before a computer can generate video from text input.
By the time that digital marketing software reaches completely automated creative production, there will be no jobs left for today’s designers, social media writers, or video producers.
Once software is capable of taking away many digital creative jobs in entertainment, advertising, or politics, what’s to stop the companies that created it from creating software that can write the next best selling novel (33 years — says that same study) or produce the next top 40 hit (11 years)?
+Level 6 — (AI)
There’s no analogous level for vehicles, thankfully, but this level could exist for a blip if any of Nick Bostrom’s thoughts about a superintelligent AI come true. In theory, an AI would create digital content from concepts to execution to launch to achieve it’s own objectives on it’s way to total galaxy domination. If we reach this level of AI, we will have much bigger things to worry about than retraining creative staff.
Future creative endeavors
While I see a coming reduction in the number of employees it will take to produce digital marketing campaigns in the next decade, not all jobs will go away. I would anticipate entertainment heavy weights, large political organizations, and some advocacy groups (the ones with significant funding) to join corporate digital frontrunners and their agencies in testing and refining the software that will produce the massive campaigns of the future.
But small agencies will be slower to implement automated creative tools. Much like there are still talented humans who hand-paint signs, there will also be talented designers who can design something more beautiful and more original than what software might produce. There are two keys to being employed in such a capacity — your work has to perform better than software-generated creative work and you have to be able to justify your costs.
For the many of us who have enjoyed the social perks of being that friend who is good at Photoshop, exchanging a random t-shirt design for some other favor in return, our days of technical and creative superiority are numbered. The kids coming out of the the “good design schools” might understand the nuances of typography or principles of design better than the kids coming out of online bootcamps, but their portfolios are starting to look more and more alike.
Creative talent must recognize that all the software that has helped them get better, faster, and produce more work will soon take away their ability to get a job. The production tricks provided by software is closing the gap between mastery of execution skills and good taste more quickly than ever. But fear not, original thinking hasn’t been programmed. Not yet, anyway.