The automation revolution is in full swing.
It began by replacing humans who were doing menial, repetitive jobs. Goodbye, factory workers.
But today, we are seeing the automation of what we thought were ‘smart’ jobs.
Take journalism. Robots will never be able to get drunk in the afternoon with their sources, of course. But they can write headlines. Algorithmic journalism is already changing the game, according to this New York Times article, which reveals that approximately a third of the content published by Bloomberg News is already created using automated technology.
Lawyers are next. The legal world are getting their briefs in a twist with news of AI startups such as Ailira and ROSS.
In a recent challenge which involved spotting errors in NDAs, twenty top American lawyers were comprehensively outperformed by an AI called LawGeex, which achieved a 94% accuracy rate, compared to the lawyers’ average 85%. And whereas the humans completed the test in an average of 92 minutes, the AI took just 26 seconds.
It turns out that just because a job is smart, doesn’t make it protected.
A report by McKinsey posited that “currently demonstrated technologies could automate 45 percent of the activities people are paid to perform.” McKinsey rated the most vulnerable professions as ones requiring rote work, data collection and processing, and some repetition.
If you think about it, that description could apply to many supposedly ‘high-level’ jobs, such as investment bankers and surgeons. After all, what is a doctor but a person who has swallowed a giant medical flowchart?
So what about creative jobs? Is it possible to automate the activity that we think of as, perhaps above all others, uniquely human?
Rice University computer scientist Moshe Vardi told the American Association for the Advancement of Science: “We are approaching a time when machines will be able to outperform humans at almost any task.”
Any task.
So will there soon be a robot sitting in the director’s chair? Coming up with advertising concepts? Producing shoots?
In some areas, there already is.
Basic editing functions, for example, can already be taken care of algorithmically. Google’s Top Shot feature, available on the Pixel 3 camera, takes a long series of photos for you and then automatically selects the best one. In short, it chooses a better photo than the one you actually took.
And AI has already been used to replace creatives, in executions that have even been awarded Grand Prix at Cannes.
However, if you delve a little closer into these ad campaigns, they’re not as impressive as they first appear.
Let’s look at three examples:
Firstly, ‘AI CD Beta’. This was a project created by McCann Erickson Japan to come up with an ad campaign for Mondelez’s Clorets Mint Tabs. The AI CD came up with the following script:
“Convey wild with a song, in an urban tone, leaving an image of refreshment, with a feeling of liberation.”
The client generously described it as “ambiguous”, but ‘nonsensical’ might be more accurate.
The exercise generated a lot of PR, but truthfully, this was nothing more than a stunt.
Second example – JWT Amsterdam got a computer to paint a new Rembrandt, for an ad campaign that won 16 Lions including Grands Prix in Cyber and Creative Data.
Except it didn’t really paint the next Rembrandt. It just painted a composite of all the previous ones.
And although it’s true that a computer physically executed the painting, who had the idea? A human.
A third example of note came from M&C Saatchi in London, who created a billboard for a fictional coffee brand, and fitted it with facial recognition software. The billboard measured how long people spent looking at the ad, and adjusted it accordingly. When people paid less attention, it varied elements such as the image, the headline, or the typeface, in an attempt to come up with a combination that people would find more engaging.
Essentially, they invented programmatic,
but for out-of-home. And like programmatic, these ads suffered from one major
problem: they were shit.
The issue, of course, is that computers are only as good as their inputs.
The Japanese ‘AI CD’ that created the Clorets Mint Tab commercial, for example, operated from a brief inputted by the ad agency planner, solo, in his air-conditioned office. But real briefs can’t work that way. Real briefs are the product of endless negotiation between the client, the agency, consumers (via research), the chairman, often the chairman’s wife, and numerous other stakeholders.
And the AI CD wrote the script from inputs too – it was fed a database of one thousand award-winning commercials. Now, it’s possible that it could have come up with something great from this soup. After all, in the words of Steve Jobs, creativity is just “connecting things” and it could have forged an interesting connection and come up with something new and creative.
But even a thousand commercials is an insanely limited and inward-looking collection for a database. The database that a real creative uses, aka his or her brain, contains many more than a thousand ads. It contains a lifetime of them. And it also contains a lot more than just ads. Real creatives are combining ideas from a potent mixture of not only all the ads they’ve ever seen, but all the movies, music videos, artworks… and all the random conversations they’ve ever had in a pub.
As for replacing directors… you’re talking about replacing alchemy. Directors are hired for qualities like sensitivity, vision, tact, drive, imagination, judgement, and improvisation. None of which are on the list of qualities McKinsey believes computers can replicate.
However, the real answer to the question of whether AI can ever replace creative and directors is not, if you’ll pardon the pun, a binary one. It may be that computers don’t replace directors, but work alongside them, as assistants.
Computers could, for example, quickly source imagery for directors to use as reference. They could pinpoint the differences between two alternative edits. And even help resolve, via eye-tracking and neuroscience technology, questions of how an audience is reacting to a scene.
The future, or at least the immediate future, could involve creative people working together with robots, not in competition.
In the words of Justin Brown, a partner at Brown Brothers Law in Michigan – who took part in the humans vs AI legal challenge – “the future is ‘human and computer’ versus ‘another human and computer.’ Either working alone is inferior to the combination of both.”
The robots are coming to replace us, of course they are.
But not just yet.