But Go is no ordinary game – more complex than chess, it reportedly has more possible combinations of moves than there are atoms in the visible universe.
Despite the game’s complexity, a Go-playing AI has just reached a huge milestone. AlphaGo, created by London-based Google subsidiary DeepMind, has won a three-game match against the world’s number one Go player, 19-year-old Ke Jie. The Chinese player hails the AI as “like a God” compared to its previous iteration a year before, and its designers say it could give a glimpse of how powerful and useful AI might become in future. What is the scope for its application, and could it transform the future of engineering?
“The way in which AlphaGo has unearthed new creative moves and strategies gives us a glimpse of the possibility of AI-aided scientific discovery,” says a DeepMind spokesman to Professional Engineering. “The techniques underpinning AlphaGo and much of our other work are general purpose and could potentially be applied to a wide range of other domains.
"We believe that in the next few years scientists and researchers using similar approaches will generate insights in a multitude of areas, from superconductor material design to drug discovery.”
Other uses for the rapidly developing technology could include monitoring workers as manufacturers embrace more automated robots, says Oxford University computer science expert Stephen Cameron to PE. “In a factory where you have robots near people, or some other sort of mechanical device near people, an AI which is able to spot when a person is not paying attention and could be hurt, could say ‘I don’t like this, I’m going to stop the motors,’ or something like that.”
Some companies have already embraced the monitoring ability that AI offers. Google has used DeepMind AI to cut cooling electricity bills by 40% at its data centres.
Novel new approaches in engineering
The technology could also lead to more “transparent and dynamic” manufacturing processes and business, says University College London expert Peter J. Bentley to PE. “The best placed industries are those with good automation and monitoring already in place so that the AI technologies can use the data in order to optimise and improve the processes.
"Ultimately AI is about distilling large amounts of data into useful knowledge, so those industries with the best quality data will be the ones to benefit in the future.”
AI’s ability to simultaneously process vast amounts of data from disparate sources will allow it to take novel new approaches in engineering, says Oxford University AI expert Stuart Armstrong to PE.
Engineering today tends to be broken down into separate concepts that humans understand, he said. Giving the example of building a bridge, he said engineers start with the concept, before considering aspects such as basic structural issues then tension, compression, resonance and other aspects.
However, an AI could consider all data simultaneously and produce a solution unlike anything created by humans, Armstrong says. “Maybe an AI of comparable abilities, but with the capacity of integrating all sorts of different knowledge, could construct something that humans don’t really understand, because it’s not in separate human-sized chunks.”
Difficult to make predictions
Despite agreeing on some definite uses for AI, such as the burgeoning autonomous vehicle sector – which could optimise manufacturing supply chains and introduce other technologies to the industry – many of the experts who spoke to PE said one thing: it is almost impossible to even comprehend most of AI’s potential uses.
“It is really difficult to make predictions for a whole bunch of reasons,” says Lancaster University’s Ben Wohl to PE. “We have advancements in a whole bunch of areas so the combinations are sometimes hard to forecast.
"But then the other thing is the nature of Moore’s Law [which projects an accelerated rate of computer improvements] and technological advancements. It is really sped up to an incredible pace at the moment, it is hard to use past experience.”
There will be applications of AI that we haven’t even considered, he said. “I think we have to see AI within a broader kind of advancing digital economy which includes intelligent algorithms alongside digital manufacturing and localised manufacturing.”