Chess computers are a good example.
“For early chess computers, humans could always win by sacrificing material for other kinds of positional advantage that the computer could not quantify but the human could intuit… computers had always been notorious for succumbing to the sins of materialism and greed”. (Joseph Ganem | The Robot Factory)
In 2001 – a film made in 1961 – Kubric and Arthur C Clarke’s imagined AI HAL (named after the three letters alphabetically preceding IBM) reversed this real-world constraint, by playing a queen sacrifice to crush his human opponent, Captain Poole.
The move underlined HAL’s capacity for long-term, strategic thinking. A capacity it later used to sinister effect.
In 1997, the real world caught up with 2001
Chess-programme Deep Blue lined up against the then greatest chess player of all time, Gary Kasparov.
As Joseph Ganem recalls, “Prior to the match, many predicted that Kasparov would win easily by playing anti-computer chess”, leveraging those clear advantages that human intuition and ‘context-appreciation’ have over machines… don’t they?
Deep Blue bested the best that humanity could offer, and to many, the trajectory of AI started to hit home.
Fast-forward to 2017
AlphaZero, an engine built on a new machine learning system taught itself to play chess in four hours, and promptly thrashed the then world’s-best chess computer, Stockfish in a 100 game match…. Without losing a single one. If you’re into chess… you got to witness what it would be like if super-intelligent aliens got hold of the game and crafted a previously undreamt-of, superior way to play it.
We may not need to worry about the pod bay doors yet, but it seems pretty reasonable to worry about how long we mere humans will remain useful when it comes to performing any occupation based on a series of decisions and inputs (…which pretty much sounds like all occupations).
So what next?
In the end, the robots are coming, inevitably. The only question is: what room does that leave for us?
What role does it leave us, specifically, as PPC marketers?
Our refuge in the ‘art’ side of PPC won’t last long… Google Ads Express already composes passably decent and effective ad text. Increasing levels of natural-language sophistication combined with the sheer amount of data available for testing and optimising will see AI-based writing surpass our efforts there in the foreseeable future.
Where will we go then?
We may be pushed higher up the food chain rather than down it… at first.
For a while, it will take a human to understand how and where to apply the new AI tools / what they are and aren’t capable of / how much budget to give them… and to explain and advise on all of this to those who don’t understand the tools but stand to gain from them.
In a sense this isn’t so different from the way PPC has always worked. How much credit can we really take for applying a bid adjustment? All we do at the implementation level is click – type – click. The ‘real’ work is then done by a series of inscrutable machines and algorithms. Our value as marketers is in knowing when, where and to what degree to make that adjustment.
But over the next few years, our useful input will be ever less at the level of these individual adjustments, and increasingly in the bigger picture. Less tactics; more strategy.
As we move higher up the managerial hierarchy, we will of course be pursued by the AI… which will become ever more capable of understanding bigger-picture, nuanced and complex requirements, and meeting them.
So our best defence is to think bigger.
We can improve our longevity by becoming more adept at this higher level, strategic side of our work, with an understanding of marketing across channels, and a sensitivity to the big-picture context and requirements of our clients.
It’s probably no longer controversial to accept that AI will become more competent than us at most tasks at some point.. and it’s clear that as the AI tide rises, the higher ground will be with those bigger-picture, abstract, less predictable and more socially nuanced areas.
The less certain – and more controversial – question for us is how perfectly the goals of the AI will match our own.
Automation already available in Google Ads often gives the appearance of (for example) favouring quantity over quality in a way that doesn’t work out well for the advertiser… Automated targeting on the GDN; automated recommendations, and most flavours of smart bidding are all often guilty of this.
Our hope has to be that as AI improves functionally, its output will better match our goals… but the truth is that Google has its own, legitimate aims that won’t always align 100% with those of its advertisers, and Google’s AI will pursue them ever more effectively.
Still – while the algorithms may have elements of someone else’s agenda – to move onto the higher ground as the AI tide rises, we will need to see them not as our competitors but as our workforce… at least until they come knocking on the office door… or lock the pod bay one.