It’s not technology that inspires – it’s the way humans respond to it

Watching the AlphaGo documentary on Netflix shows how marketers can elevate themselves through AI –and now’s the time to start

July 8, 2019

It’s not technology that inspires – it’s the way humans respond to it

My Netflix account categorises the film AlphaGo as ‘Inspiring’. For about two-thirds of the movie though, I wasn’t sure. Artificial Intelligence (AI) is something I’m fascinated and excited by. The way I see it, human beings have been trying to create technology that can transform our abilities since the beginning of time. We’ve designed tools to make tasks easier, quicker and more efficient – and to achieve things that would have been unimaginable without them. AI is the latest and potentially most thrilling step in this process. However, it also generates the most fear. And AlphaGo helps to show why.

This documentary reveals how DeepMind’s AI team spent two years building an Artificial Intelligence (AI) system to crush one of the world’s greatest players of the world’s most complex game. It explains how AlphaGo’s three layers of neural networks interact to replicate the workings of the human brain; how it can ‘see’ so many more different outcomes, possibilities and courses of action in the game of Go than even a 9-Dan Go master. But is something being lost in the process? Is DeepMind’s team of computer scientists taking something ancient, mysterious, contemplative and beautiful – a game that is almost a form of meditation – and putting it out of people’s reach? Is Go still the same game if a machine can almost always play it better than a human can?

As marketers, we face similar dilemmas. We’re very aware of the competitive advantages that AI has to offer. We know that we can’t afford to be left behind – and we’re interested in any opportunity to apply AI to different areas of strategy. However, there are inevitable misgivings too. When AIs like AlphaGo are able to turn even the most remarkable human brains into just something to prove their superiority over, what chance is there of our ideas and creative contributions being valued over those of an AI system? Are we relegating ourselves to a supporting role in the future of our profession?

Why AlphaGo should inspire marketers
I don’t believe we are. And the reasons why are perfectly captured in the final third of AlphaGo. This is a film that’s fundamentally optimistic – and surprisingly human. It inspired me about what our future relationship with technology can be – and I hope it inspires you too. Because there are already plenty of opportunities to apply that technology in the way that we approach business and marketing today. These applications aren’t the kinds of things you’d create a Netflix movie about. However, they are still applications that can make us far more than we would be without them – provided we respond to them in the right way.

In AlphaGo’s tale of human vs machine, the heroes aren’t just the programmers and computer scientists. Even more heroic are the two Go players that their machine beats during the course of the story. In their different ways, they show how human experience, creativity and empathy will never be redundant when it comes to getting the most out of AI. Even better, they show how AI can unlock people’s potential – helping us to become more, not less.

Fan Hui and the importance of interrogating how AIs think
The first of these players is Fan Hui, a European Go champion and a 2-Dan player in a ranking system that goes up to 9-Dan. Fan Hui is very confident about beating AlphaGo when the DeepMind team invite him to play a match against the machine. Instead, he’s crushed by a piece of technology that’s far beyond the Go-playing simulators he’s encountered before. This affects him deeply. His sense of self-worth is threatened. However, he rebounds to help the DeepMind team overcome a fundamental problem with any machine learning system.

At this point in the film, AlphaGo knows far more about Go than any of its creators do. It’s therefore impossible for them to know how good it really is. Is it reaching new levels of ability – or is it coming up with flawed ways of playing that will quickly be exposed by a true Go master? When Fan Hui looks for weaknesses in AlphaGo, he discovers that the system’s deep learning is actually very patchy. There are areas of the game that it’s understood brilliantly – but there are others that it’s effectively ignored. Without the skill and imagination of a human expert, AlphaGo would have ended up limiting its own capabilities, with none of its creators being any the wiser.

As AI systems become increasingly advanced, and do more of their learning themselves, this type of human intervention will become increasingly important. Without a Fan Hui to challenge it, there’s no accountability for an AI system – and no guarantee that it is really producing the best possible results. On the other hand, Fan Hui himself seems transformed by being reconnected to the AlphaGo project. His skill levels are looked down on by other Go professionals earlier in the film, but by working with the machine he is taking the game itself to another level. By the way he rises to the AlphaGo challenge, Fan Hui transforms the impact that his own skill and passion can have.

When humans respond to the challenge of AI
The ultimate test that Fan Hui helps to prepare AlphaGo for is a five-match series against a 9-Dan Go master: Lee Sedol, one of the greatest players in the world. Like Fan Hui, Lee Sedol is supremely confident in his ability to beat the machine – and as with Fan Hui, that confidence turns out to be badly misplaced. AlphaGo defeats him in the first three games, settling the series at the earliest possible opportunity. The expert commentators fronting global TV coverage are stunned. The Go world is in shock. And Lee Sedol? A gentle, dignified man who’s been elevated from his background in rural Korea by his incredible gift for the game? He seems completely bewildered, shocked, almost to be falling apart on-screen. It feels cruel. It’s painful to watch.

Then, though, something truly extraordinary happens. Lee Sedol searches deep within himself. He replays the games with friends and fellow experts. He somehow finds a way to get under AlphaGo’s skin. He wins the fourth game in the series by making a move that is so unprecedented it becomes instantly legendary in the world of Go: the ‘God’ move, as commentators refer to it. AlphaGo itself has calculated the probability of a human being able to make it as 10,000 to one.

By the end of the film, the game of Go hasn’t been reduced to just another thing that machines can do better than people. Instead, it’s been expanded. As one expert puts it, people have been playing this game in a particular way for thousands of years. AlphaGo has revealed whole new ways to think about and play it.

Something similar has happened to Lee Sedol himself. In the course of just a few days, he’s been able to rise to this unprecedented challenge and somehow achieve a deeper level of understanding of a game he’s played his entire life. He remained undefeated at Go for years afterwards – arguably as a result of the leap forward he was able to make. However, his dignity, courage, commitment and creativity under pressure achieve something else just as significant. The popularity unleashed by this highly publicised contest reportedly led to a shortage of Go boards worldwide. In my view, those new players weren’t inspired by AlphaGo itself – but by the spectacle of the man straining every sinew of his mind to compete with it.

Why AlphaGo’s opponents are models for marketers
I would argue that Lee Sedol and Fan Hui both provide role models for how marketers, and any other professionals, should aim to work with AI. In fact, they remind me very much of the ideas in one of my favourite books at the moment, Hello World: How to be Human in the Age of the Machine.

The author of that book, the mathematician Hannah Fry, points out both the huge advances and potential flaws in a world run by algorithms. Her argument is that AI can be a powerful force for good in the world – but only when working together with humans. If we put too much trust in something we don’t understand, leave it to make all the decisions and no longer push ourselves, the results could be disastrous.

The two Go players in this film show how this can work. On the one hand, they demonstrate how human ingenuity, craft and skill is essential for keeping AI systems honest; to ensuring that the answers they come up with actually fit the questions being asked; that they are not missing anything important. They also show how even the most advanced and intimidating AI systems don’t make human thought and human creativity redundant. They open up new possibilities – and these possibilities can be explored most creatively by human minds.

DeepMind spent two years developing an AI system that could beat Lee Sedol – one that he’d never encountered before and had no real way of preparing himself for. It then took Lee Sedol days to find a previously unimaginable way to beat that AI at its own game. That’s breathtakingly quicksilver creativity. Machines can’t compete with such speed and flexibility. Their value comes in provoking and inspiring it.

Let’s not take AI insight for granted
As marketers in 2019, we’re operating at a time when the availability of AI-driven insight is growing extremely rapidly. It’s easy to take it for granted, even when the predictions we’re being given would have seemed magical just a few years ago. IBM’s Watson-derived systems can predict which emotive words in an email subject line can increase response rates. It can even generate different versions of the subject line for you. Among other things, Amazon’s Rekognition technology can quantify the different creative elements in a video campaign: the types of characters, use of light, different visual elements, and how these different elements combine to capture human attention. On LinkedIn, the new Objective-Based Advertising interface is able to predict the results your campaign will generate – and the amount it will cost – based on the way that you define your target audience. That’s really, really impressive – and really, really useful.

The challenge for marketers is to make sure that we respond to these insights in the way that Fan Hui or Lee Sedol would. It’s pointless and depressing to think of ourselves in competition with machines. We need to take their outputs as a starting point – and apply our imagination to making maximum use of them.

With Objective-Based Advertising on LinkedIn, that can mean marketers pushing the boundaries and finding new, creative ways to target an audience. If IBM Watson suggests email subject lines for you, then use those suggestions as a starting point – and strive to find alternatives that might work just as well, for reasons an AI wouldn’t consider. Once AIs like Amazon’s map out more of the science of how people respond to video, we’ll have a choice: follow the obvious formula it suggests – or twist it in new directions and come up with something more creative, more unexpected, more game-changing.

Just as with AlphaGo, it’s the technology that sets the foundation for something inspiring to happen. But it’s the human who inspires.