The first time I played chess was in a dusky room huddled over an aged wooden board. I was 5. My grandfather was furtively explaining each piece to me, each move, the key positions and the basic tactics. That’s about all I remember from that experience. Eventually, after many years, I learned how to properly play the game, well enough to even beat him on most of our encounters – but at the time I remember being dumbfounded. Once I learned the basics, the honest highlight of each match was getting to mindlessly shift around pieces while my young interest was far more captivated by the TV playing over my grandfather’s shoulder. He absolutely destroyed me each and every time (and now that I think about it, what kind of grandfather doesn’t let a 5 year old win once in a while?), but I didn’t mind a bit – I always got a slice of cake as a compromise for my defeat.
I suppose my point is that the human brain is great at some things, and terrible at others. Recently Google announced that their AlphaZero artificial intelligence supercomputer learned chess…after four hours. Four HOURS. And it basically did it by itself. The Google team only programmed it with the basic rules, and it had to glean the infinitely complex game strategy on its own, armed with a vast trove of past matches to study and scrutinize. And learn it did – by all accounts, it could very well be the best chess mind the world has ever seen.
It not only obliterated all human competition (that’s less of surprise than it sounds – computers in general have far surpassed the chess ability of humans in recent years), it handily beat the very best computer chess programs. Just to be clear: computer chess is not a juvenile field. Programmers and chess experts have been constantly adapting, revising, improving computer chess algorithms for the better part of the past forty years, ingraining ancient knowledge and sacred wisdom into the recesses of 1’s and 0’s, bits and bytes that live cold lonely lives on a titanium microchip. Google has flipped that all on its head with a chess computer that can teach itself in four hours.
The moral of the story? AI will soon be able to do things that marketers have always done, and better. But they won’t be able to do EVERYTHING for a long time yet. AlphaZero isn’t even a chess computer – that’s probably the most incredible thing about all this. Someone simply had the idea to apply its processing power to chess and see how it fared. At the end of the day, AlphaZero is just a tool – the real brains will always be the humans behind the curtain.
Marketers must work harder than ever to hone their creativity when working with these machines, constantly innovating new uses for their ever progressing abilities. Artificial Intelligences go one step further than traditional computers: they try to emulate the human brain instead of just carry out commands and instructions. But I believe the technology has a long way to go before it can mimic genuine human originality. In the meantime, marketers will have to become proficient at applying artificial intelligences in the limited scope the current technology affords us:
- Facilitating genuine customer interactions for better customer service and support
- Mining through hoards of data and analytics to present the best insights to human eyes for further decision making
- Using image processing to organically react to User Generated Content (a time-consuming prospect for humans)
These are just a few examples of what the current state of artificial intelligence can accomplish (it can certainly do a whole lot more). One day, in the future, AI’s WILL be able to strategize the next big ad campaign; they WILL be able to write catchy copy and design beautiful billboards; they one day might even replace human CEO’s at the top of the organizational hierarchy, making cold, calculated decisions the way even the bluntest of our leaders never could. But in the meantime, we must acknowledge their limitations and use them like the tool that they are.
Just don’t expect a friendly afternoon chess match.
Senior Strategist, Snipp.