Science fiction, fantasy, computer games, and chess. My children are all over these, playing hard and imagining new worlds. As we start to face adult responsibilities, it’s sometimes hard to hold on to the absolute joy of play, but discussions around the development of artificial intelligence at this year’s South by South West (SXSW) festival reminded me how important it is not to grow up.
Playing machines at their own game
Presenting on Ethics and Responsibility in Artificial Intelligence, Garry Kasparov fondly recalled winning the world Chess championship in 1985 when “machines were weak and my hair was strong”. A decade on, in 1996, he lost the first of three matches against IBM’s Deep Blue, winning the other two: he found this even competition more interesting than “the brain’s last stand” in 1997 when the machine finally gained the upper hand. This success did not imply that Deep Blue thought like a human, or even that it memorised all the 1046 possible moves. It simply had to make fewer mistakes. This is a key aspect of the machine learning and artificial intelligence which complements our human performance. Machines do not grandstand. A win is a win. For decision making to be both precise and creative, humans and machines need to work together. A great analogy can be found in the world of motor racing. A Formula One car in the hands of a commuter may terrify the driver and will not set any lap records. An F1 star ‘in a reasonably priced car’ may be faster than the average celebrity but would not win the championship. Put the high performing machine with the high performing human, and you have a winning formula.
Since Deep Blue, technology has moved rapidly. We now have Deep Mind’s AlphaGo Zero, learning a far more complex game from scratch. According to Accenture’s UK AI lead Fernando Lucini, its ability to be curious has enabled it to deploy successful strategies which would simply not occur to a human player. How has artificial intelligence advanced so much in two decades? Thanks are due to the gaming community, said IBM Watson IoT’s Chief Agitator John Cohn. As computer games grew in complexity, hardware ran to catch up. Anyone who has gone back to the original Lara Croft games will see the difference between graphics processors in the Deep Blue era and the way things look now. I recently retired my original Playstation and its pixelated Crash Bandicoot in favour of the latest remastered, crystal clear graphics. The advance in GPUs for better game experiences has delivered the processing power needed for the new generation of artificial intelligence. Gaming has driven innovation, and to continue advancing, we must continue to play.
Playfulness in interaction
The Genuine People Personalities conferred by Douglas Adams on Marvin and the Heart of Gold spaceship doors were a hilarious warning to the builders of today’s AI. As Slack’s Anna Pickard reminded us, there is a difference between a ‘fun’ personality, and the persona we take away from any encounter. Whenever we interact with an AI (and believe me, you do this more than you think), the register and content of communication are important. Experiments with AI which mirrors emotion have not been a roaring success. Mirroring stress, for example, built a spiral of worry instead of alleviating the condition. Playfulness is a much easier way to connect, with appropriate adjustments for culture. Personalisation and trust come from friendly interaction, although this does not mean the machine is a friend. Is Alexa your friend? Do you want it to be? Ultimately, the goal is for AI to speak to you in the right way, at the right time, and for you to be comfortable enough that it empowers you to take the right actions.
Fiction and hope
For lifelong scifi fans, the fictional inspiration behind some of today’s everyday tools is not lost on us. It is more than two decades since I owned my first Star Trek communicator (Motorola flip phone) and it’s a sure bet that advances in the use of lasers are being driven by scientists and engineers who really, really want a light saber. There is a dialogue between writers and developers driving inspiration and innovation in both directions. Author Ytasha Womack led a lively discussion with representatives from Boeing, the Aerospace Industries Association, and the Smithsonian. Fiction allows for metaphors and thought experiments, and for the presentation of dystopian visions which reflect the lessons of the past or scenarios to avoid in the future. Blade Runner’s battle for replicant rights is a fictional representation of the ongoing battle for human rights. However, in the same film we can see that extrapolation of the future starts with what we already know. At the time, computers filled large rooms, and communication happened in telephone booths, and you can see these in the film. What lies in our future reality? One thing of which we can be certain is that fiction helps creative engineers to push the envelope, investing time and effort in making imagination a reality. It is the responsibility of writers to stoke the fire with stories, in what is a golden age for popular culture.
The overarching lesson on innovation is that both humans and machines have to let their curiosity lead them to new things. Be curious about your world. Be curious about your future. Be curious about the potential for machines to complement your skills and deliver better results thanks to a true meeting of (human and silicon) minds. We don’t have to grow up, after all.