To the average person, the billboard on the bus stop on London's Oxford Street was a standard coffee-brand ad.
Every few seconds, the digital poster would change. Sometimes, it would feature a wide range of drab grays and blocks of text. Other times, it was a minimalistic image with a short saying.
What was unique about this particular poster, which ran in two locations at the end of July 2015, wasn't the fact that people were looking at it. Rather, it was looking at them — and learning. Using facial tracking technology and genetics-based algorithms, the poster took the aspects that people looked at the longest and then incorporated that into the next design evolution.
"We were surprised how quickly it learned," said Sam Ellis, business director of innovation at M&C Saatchi. "It got to a state of where it felt like it was in the right place a bit faster than we thought."
In less than 72 hours, the M&C Saatchi advertisement was creating posters in line with the current best practices in the advertising industry, which had been developed over decades of human trial and error like realizing three to five word slogans work best.
"We thought [our employees] would be nervous about it: Is this going to kill off creative?" Ellis said. "What they started to realize is that it could be really, really useful based on its insight."
M&C Saatchi's Ellis believes eventually ad agencies will be smaller, because AI will be able to accomplish tasks with a high degree of accuracy — for much less money than now — and will make outsourcing tasks a lot more effective.
As our machines become more sophisticated and more details about our lives are recorded as data points, AI is getting to the point where it knows a tremendous amount about humans. It can tell what a person is feeling. It knows the difference between a truth and a lie. It can go through millions of permutations in a second, coming up with more variations than a human could think of. It knows your daily routine, like when you're most likely going to want a cold beer on a hot summer day.
"The AI economy is going to be the most disruptive thing we've ever seen," said Winston Binch, chief digital officer of agency Deutsch North America. "It's tough because the pace of technological change is so fast, we don't know what to do with it."
This leads to the question: How far are we from the day where AI is going to be better at creating advertising than the creators themselves?
"We all have to be a little unnerved by this," Binch said.
A $10 BILLION INDUSTRY AND GROWING
Some forecasts have pegged the AI industry at $10 billion globally today, with spending growing in the double digits annually, according to Deloitte LLP. A United Nations report suggested as much as two-thirds of all jobs in developing countries could be affected by "significant automation."
"There is no question that there is strong and rapidly growing investment in this area, spanning hardware, software, cloud and services," said David Schatsky, managing director of Deloitte LLP. "But the impact of cognitive will likely be much greater than the sum of the investments in the technology. A growing number of companies are already benefiting from cognitive technologies that are embedded in the products and services they buy and use."
Businesses like Netflix use AI to power its movie and TV show recommendations, according to a Deloitte report titled "Cognitive technologies: The real opportunities for business." Meanwhile, the Associated Press uses language-generation AI to write corporate earnings stories.
The concept of machines being capable of logical reasoning has been around since the 14th century, but it wasn't until 1956 after a conference at Dartmouth College that people began earnestly researching the subject. By the late 1990s, AI was being used for data mining and other logical deduction tasks.
But, it wasn't until March of 2016 that AI was able to do the previously impossible: Beat a human at a game that required highly advanced reasoning and deductive skills. In other words, Google's AlphaGo was able to "think."
Go is a Chinese game that involves using stones to take over more "territory" on the playing square. The number of possible moves in the game is said to be more than the number of atoms in the observable universe. The human opponent had some advantages over AlphaGo: The computer couldn't shift its playing strategy once the match started, or think on the fly and be creative with its moves.
That's because, for now at least, the best AI only has about the equivalent of a couple hundred brain neurons, said Stephen Pratt, CEO of enterprise AI company Noodle.ai, a partnership with TPG Growth. For comparison, the human brain has 100 billion.
"For the foreseeable future, you absolutely need humans as the genesis of ideas and the creative director, and to understand the final product is good and it is effective," Pratt said.
In the movie "Minority Report," "precogs" were able to determine if someone was likely to commit a crime, Pratt said. In advertising, while there aren't mutated humans capable of determining your future, AI can "absolutely" determine if a person is more likely to purchase a product and what kinds of things might compel them to want to buy it. It's not quite the same thing, but the technology is getting there.
"In business, all you need is a slightly increased probability to be effective," Pratt said.
Currently, the advertising industry utilizes AI in two basic ways: to analyze people through data and machine learning and to create technology-enabled interfaces to complete basic tasks, said Josh Sutton, Publicis.Sapient's head of artificial intelligence. Many agencies are launching AI divisions, mining the data they have on users through ad tech firms and social media platforms to create advertising faster and, theoretically, better.
"[AI] not being creative per se," Sutton said. "It's understanding that 'if I show these kind of things to a person, I can evoke this kind of response.' I hate to say it, but it's rules-based — but a lot of our emotions are rules-based."
AI GOES BIBLICAL
To promote Canal+'s series "The Young Pope" in France, the show teamed up with agency Havas and IBM Watson to create AiMen. Its mission: "What if in the XXIst century a Pope decided to use Artificial Intelligence to preach God's word to Internet users?"
The team behind AiMen uploaded all their research and scripts into the program. From there, the AI parsed out tweets, Facebook posts, YouTube or Daily Motion comments that had expressed anger, envy or egotism. Then, the AI would respond with a Biblical verse. If the user responded, the AI would continue its pious banter.
"Eeeeeeeeee I become really mad when I see a GT86, it's such a beautiful car :(" one user tweeted in French.
"Better the little that the righteous have than the wealth of many wicked; Psalms 37:16," the AI tweeted back.
"No, but Lyon is really a losing team," wrote a Facebook user. "Not f****g winning..."
"Repent of this wickedness and pray to the Lord in the hope that he may forgive you for having such a thought in your heart; Acts 8:22" the AI responded.
With each interaction, AiMen learned to be more clever and witty. Since it launched on Oct. 28, it has analyzed upward of 4.45 million posts, responded back almost 885,000 times and learned a little under 40,000 Bible verses.
"I think it's the first time we created an intelligent agent, acting as a marketing promotional tool for a god object," said Jason Jercinovic, global head of marking innovation at Havas. "Our thesis is some people are not going to know this is a marketing campaign. They're going to go to battle with the bot, and that's interesting."
While it can interact and "read" emotion, AiMen still needed humans to create the content it used for "inspiration." It can't create its own moral code and Bible from scratch, nor could it create a personality without the help of the team that created the show. But using those references, it could discern what emotion people were expressing and respond all on its own.
"The promise of what AI will be able to do is to start to knock on the door with emotional words or knock on the door of what will lead to do emotional decisions, said Alan Schulman, managing director of brand and creative content marketing for Deloitte Digital. "But the end of the day conception or ideation is a human thing."
Advertising is such an emotion-based industry, Schulman pointed out.
"Algorithms don't feel," he said. "People feel."
What AI is leading to is the personalization of marketing, a concept of creating "markets of one," Havas' Jercinovic explained. Instead of a billboard that is likely to appeal to the masses, marketers can create personalized ads that will appeal to you. And, brands can reach you on all your devices, whether that's on TV, your work computer or your phone.
AI will also take over the more menial, basic entry level tasks, he said. Already, on a basic level, he pointed out we rely on AI somewhat to schedule our meetings or book a flight with a chat bot. We yell for Alexa if we want to know movie times. Siri can send texts for us.
"I see the tools and platforms that are coming out that are AI enabled assisting the doctors, not replacing them," Jercinovic said.
"What does it mean for the people that went through that to become the partners in the law firms or the nurses because they were the doctor's aides for a long time?" he pondered. "That's what the question is."
WHEN MACHINES BECOME MORE THAN MAN
Jercinovic, however, isn't worried about creative industries like advertising. There will always be a need for people to come up with the idea for an advertising campaign or write (or at least approve) clever text to sell a product, he said.
"We as creative people want to be involved in that process," Jercinovic said. "On the recipient side of it, we're still making sure that voice is right and that experience comes to life."
Media planners who decided when and where to buy advertising may find their jobs automated because of AI's ability to perfect the right time for supply and demand. M&C Saatchi's Ellis admits, however, that the technology is far from that day.
Expect both changing roles in the industry — like copywriters working with AI — and new jobs to emerge managing AI platforms, Ellis said.
AI has lead to the integration of creative and technology teams, Publicis.Sapient's Sutton pointed out. At the end of the day, while some jobs may no longer be necessary, he feels that there will be an increase in the number of people needed to complete these more complex, more personalized ads. There may even been needs for new positions, such as executives with AI specialties.
"Every time there has been a big new technology, there has been a groundswell of people asking is this going to replace all our jobs?" Sutton said. "Is it going to be chaos? What consistently happens is there's more work created or more work needed to meet the new demands."
What is clear is AI will help us create new or unique ad formats, said Deutsch's Binch calling it a "liberating force." One of his projects, Volkswagen's Unleash your Rrrr, is an artificial intelligence-enabled program that lets people control a virtual car around a winding road by making car sounds. Rrrrs, vrooms, eerrks and the like let cars accelerate, turn and screech along hairpin turns. It was the first AI that was capable of understanding abstract sounds and emotions.
"Writers have a place in this new AI economy," Binch said. "There's a new place for technologists. It reformats us in a way that we may not be coming up with a technology solution [for our advertising jobs right now], but we're going to be able to use this technology in new and creative ways on behalf of a brand."
It's allowing people to focus on the creative side of coming up with ideas, rather than the nuts and bolts of engineering the advertising campaign.
"AI is going to replace everybody, but we're a ways from that," he added somewhat reassuringly. "AI can do so much, but it can't replace human charisma or creativity — yet."
More from Future of Work:
How software is eating the banking industry
Lawyers could be the next profession to be replaced by computers