Artificial intelligence and robotics: The end of the world (as we know it)?
14 April 2015
There have been a lot of developments in robotics in recent years, with a focus on artificial intelligence and machine learning. More recently, there have been several stories about practical uses of robots with ‘artificial intelligence’ in everyday life, from simple things such as delivering packages in a specific area, to serving customers in one of Japan’s largest banks. So, I wanted to write an article to discuss these topics and pull together a few ideas of what more we can expect from future developments in this field, and whether these developments will ultimately be a good thing.
Many of you will have heard of Amazon’s trial run of using drones to deliver packages that had been ordered online, back in 2013. Drones were programmed with a specific route and are supposed to be able to traverse to their end destination, and then return back to their starting position. This is an idea with quite a simple concept, but one that has huge practical applications and many of the world’s largest delivery companies are anxiously awaiting completion of successful delivery drones. Amazon is still testing its drones here in the UK, but a recent announcement by the Federal Aviation Administration has made delivery drones active in US air space a much more difficult achievement.
Have you ever been stuck in a long queue at the bank? If you have, now you’ll wonder if it’s because the robot serving all the customers is re-booting – yep, robots serving customers! Mitsubishi UFJ Financial
Group, Japan’s biggest bank, is employing a robot called “Nao” to assist in serving its customers. Nao has a camera on his forehead that allows him to see his customers and recognise the tone of their voice – he can then interact with his customers, greeting them and offering the bank’s services to them. He has a great knowledge of finance and can speak 19 languages. Depending on his performance, more robotic staff will be employed with the hopes that they can perform tasks that human employees can’t, such as helping when there is an increase in foreign customers and working 24 hour days.
While you could argue the convenience granted by the examples above is great, my next example is actually helpful. In Australia, robots are being developed to assist in farming. Yield loss is a real issue, and competition for minerals and nutrients by plants means that most crops suffer. The plan to introduce robots to farms should revolutionise the farming industry and increase the yield of crops, which is beneficial not only to farmers, but to all consumers.
So it seems that everybody wants to make robots. I’m not sure if it is due to watching Robot Wars too much growing up and thinking making robots is cool, or that people can see the benefits of having machines that can operate and perform tasks autonomously. One thing is for sure, currently these are all just pre-programmed machines and do not truly have artificial intelligence. In fact, I watched a film recently, called “ex-machina”. It is about artificial intelligence in robots and how a machine cannot possibly possess artificial intelligence unless it can pass the Turing test. However, in this film, a spin is put on the standard Turing test and a third party must have a conversation with a machine, knowing that it is a machine, and yet still be able to believe it can ‘think’ and process its own answers to his questions. Now back to reality. Developing a machine that can think and act for itself, as well as being indistinguishable from a human, is a long way off. The closest we are to achieve this currently is to go through a process of machine-learning. This process requires a machine to mass calculate models using example data so that it can learn from the answers it obtains – it should then be able to make informed decisions or predictions about other things, based on the knowledge it has from the data it has already seen. That is a very high level description, but more can be read here. IBM’s Watson computer is probably the pinnacle of modern machine learning and is able to understand natural language and generate hypotheses to answer questions based on evidence obtained from unstructured data. The learning works quite similarly to how a human would learn – Watson can guess, and will learn from mistakes and feedback it receives – it learns dynamically, getting smarter with repeated use and learning as it goes.
The developments in these areas are amazing. Worldwide economies will boom as robotic workers become more of a norm and gain an increased knowledge of finance and as robotic farming creates further efficiencies in farming and food shortages are a thing of the past, we will laugh and be care-free as we slowly gain weight and begin to live a life floating along on our hover chairs, looking like the humans on “Wall-e”.
But then you have to ask yourself – what happens when the machines learn? When the machines become capable of thinking for themselves and realise that we sit there incapable of doing anything for ourselves because years of having tasks automated for us has meant we no longer possess the skills or knowledge required. Will they take advantage of our weakness? Will the machines, possessing advanced financial knowledge as well as control of the supply of food and its delivery, decide to extort us so that they can further prosper from their resources? Or will they rise up in anger at how we have treated them as tools to make our lives easier, and decide to revolt and slay us all like in Terminator? Which poses the final question – which one of our leading technology firms is SkyNet, and how do we stop them?
If you would like to discuss these issues, or the impact of emerging technology on your industry, then please get in touch with Euan Cameron.