Artificial Intelligence and the Future of Advertising

Professor Stephen Hawking recently warned the world that “the development of full artificial intelligence could spell the end of the human race.” While this may bring images of a Terminator-style apocalypse to the mind, it is still farfetched based on existing technology. “Full artificial intelligence” or machines that match or surpass human intelligence do not yet exist. However, artificial intelligence (AI) working in collaboration with humans is more common now and has impacted many industries, including media and advertising. It’s exciting to be able to get a sense of the future by looking at some of the ways AI has already and will continue to impact marketers and the advertising landscape.

Perhaps the biggest impact AI has had on advertising is in digital media via programmatic buying and planning platforms that utilize complex algorithms to perform a variety of functions. Connecting ad buyers with sellers to serve individual users with ads at the optimal price can now be achieved with little or no human intervention. While bidding for an ad, these complex decision science systems also utilize machine learning to decide which ads should be served to which consumers.

Systems like this take into account multiple factors such as how often users are online, their browsing behavior, what types of ads they respond to the most and what types of ads they ignore.  To determine cost, the data informs the bid model and determines probability of success in serving an ad to the right type of consumer. The higher the probability of success, the higher the bid will be. This happens billions of times a day for thousands of advertisers, all in a fraction of a second. However, if this is what machines can do for us now, where does it go from here? In short, the answer is simply that there will be a higher volume of programmatic across all formats. Programmatic technology has the potential to encapsulate all media buying and planning across every channel, including TV, out-of-home, print and radio, but the infrastructure isn’t yet ready for this. As trade activity increases in scale, it is certain that one of the future technology giants vying for the largest volume will launch a universal planning and execution system that can merge all forms of offline and online media. Advertisers will not be able to ignore the tremendous benefits of an interface like this. What’s more, the increased cross-channel measurement and real-time insight possibilities will make ads more relevant and targeted than ever.

Last year the biggest U.S. magazine publisher, Time Inc., started selling the inventory of its print magazines programmatically. The thought behind this move was that the ability to buy print programmatically might help attract ad budgets to the medium. Although it is a far cry from the algorithmically targeted digital ads, since Time Inc.’s offer covers just 18 demographic or interest-based groups, it is still a groundbreaking effort that points out new opportunities for programmatic buying.

Katz Radio Network, one of the biggest in the U.S., has also started to sell radio spots on its programmatic exchange. The exchange does not include real-time bidding, but is more focused on automating the buying process, which is currently extremely labor intensive. However, most of the research and development in offline programmatic is happening around TV.  IPG Media Brands’ programmatic arm Cadreon, has partnered with TubeMogul to build a TV automation platform. This marks one of the first examples of a proprietary, and inventory-agnostic demand-side platform for TV. A panel of buy-side and sell-side experts at comScore’s industry summit on the future of audiences and advertising agreed that this isn’t just an isolated test; TV is indeed going programmatic. According to a recent report by IPG MediaBrands’ Magna Global, programmatic TV will reach 17 percent (or $10 billion) of global TV budgets by 2019. Reliable figures on programmatic spend for offline formats, such as magazine and radio, aren’t yet reported. However, continued growth in automated buying for those formats is inevitable and tied only to the speed at which infrastructure and systems can be upgraded.

On the creative front, AI is now being tested to develop creative assets and ad copy based on real-time learning. Last year a partnership between M&C Saatchi, Clear Channel and Posterscope created “the world’s first ever artificially intelligent posted campaign,” at least according to their words. The program reads audience reactions and changes its behavior depending on that interaction. It starts with a broad set of photos and copy, and from there, 22 ads are created for each test run, with the program determining the overall success of each ad. Ads that work then move on to the next round, and those deemed unsuccessful are removed. David Cox, chief innovation officer of M&C Saatchi told the Guardian last July, “It’s a Darwinian algorithm; it’ll evolve to be more and more effective.” As of that month, 1,540 ads had been automatically served. The first set of results showed that shorter copy seems to be most popular, and images of hearts were used more frequently.

If a machine is able to write copy and employ decision sciences to continuously improve ads in real-time, what implications does this have for the future of creative ad work? As it stands now, the human touch is still required to fine tune the final ad and the machine generated ads can be chosen and signed off on before they are served. In fact, human touch will be necessary for the foreseeable future, at least prior to the point where a machine surpasses human “intelligence” or intuition. Still, interesting examples of this increased automation are popping up in other sectors of the media business as well.

On the editorial front, the Associated Press (AP) partnered with Automated Insights and now uses their “Wordsmith” technology to publish quarterly earnings reports written entirely by this AI platform. Before this implementation, AP estimates it was doing 300 reports each quarter. Now it generates 3,000 such reports every quarter. Obviously if this kind of technology is successful and eventually made widely available to newsrooms across the country, the industry can expect staffs to be trimmed even more down the road. Of course, continued efforts by IBM with its Watson program will eventually bring many beneficial decision science services to not only media companies, but those in practically every industry imaginable.

Another big storyline that marketers will be watching in the coming months is Adobe. At Summit Europe in April 2015, Adobe announced the company acquired Tumri advertising technology to extend its programmatic advertising capabilities. The technology enables dynamic creative optimization based on website behavior, audiences and context in real-time at the last millisecond of marketing. Surely if a vendor of this size continues to make these kinds of acquisitions, it will have some wondering if Adobe might become the first big player to unite all media formats into one programmatic system.

The ability of machines to analyze vast amounts of data to learn, adapt and create has the potential to unleash insights that may have been beyond the realm of human imagination and that is why tech giants like Google, Microsoft, Facebook and Adobe are investing heavily in AI.  Advertising will undoubtedly be impacted by automation technology, and both media and creative agencies will be leading automation research, design and implementation. The job market will inevitably evolve too, but it will also liberate many from doing menial tasks, allowing them to spend more time focusing on innovation and strategy. Advertising professionals and brands alike will have to be progressively savvy with digital and comfortable learning about the inner-workings of programmatic technology since it will eventually become one of most important keys to creating successful ad campaigns.