Dr Carl Frey and Dr Michael Osborne of Oxford University recently delivered a lecture entitled: “Replaced by robots? The challenges and opportunities of automation for the workforce”, where they discussed the impact of automation over the next couple of decades. Whilst acknowledging the difficult of predicting the future, they felt that the changes that will be seen over the next 20 years are likely to be from the technology which is already in existence today.
Dr Osborne's DPhil work on sensor networks had significant influence on the EPSRC/Industry research project ALADDIN, winner of the Engineer Award Aerospace and Defence 2009, which is featured in the Institution's current
Vision book.
Gradually, over a couple of hundred years, low paid workers have been replaced by automation, allowing them to move into other areas and potentially opening up a skills gap.
Computers are very good at routine work – and they don’t need food, holidays, get sick or bored. And more importantly there is no variation in response, they make very clear headed decisions and over time they are cheaper than the human alternative. The introduction of the first industrial robots was swiftly followed by complete automation. As long as you can break down a process into a series of definable, repeatable tasks, then it is likely that it can be automated. And what’s more, the cost of automation has dropped dramatically.
In the economy as a whole, manual employment has decreased and service employment has gone up and the wealth of the middle income workers and highly skilled employees has increased.
Today the game changer is big data. Jobs that were thought to be beyond the realm of computers are suddenly well within the automation possibilities. Workers in cognitive tasks, not just those doing manual tasks, are now being replaced by automation. Analysing vast quantities of data, recognising patterns, predicting responses are now not only possible, but in some cases easy.
Translation is a good example, the EU works in 6 different languages and the translations are improving as machines are able to exploit pre translated works to identify phrases. Consider the legal profession, where sifting through masses of information, is now much better done by computer than by paralegals – algorithms are better at the job than people. You have only to make a purchase on Amazon, to be bombarded by your previous purchase information providing the basis of your recommendations for future purchases.
So where should you work to be safe? Firstly computers are no good at social intelligence, negotiation or debate. So if you are a teacher, a social worker or a barrister you will be very hard to automate. Similarly creativity is hard to replicate – so artists and musicians are safe. And finally autonomous manipulation in unstructured environments is very difficult for a computer. Complex tasks, lots of data on perception, deep understanding of objects and surrounding, are all barriers to automation. Homes are usually less structured than work environments.
Typically a wide range of objects may be scattered in different positions each day, which makes life hard for robots. What is the difference between dirt in a plant pot, compared with the dirt fallen off your shoe? Workplaces, in contrast often have flat floors, good lighting, with processes which are definable, repeatable and predictable. This makes hospitals, warehouses, and airports ideal for automation. Warehouses can have barcodes on the floor to solve navigation problems, and can therefore be fully automated. Hospitals are already introducing robots for delivering food and medicine.
The probability of computerisation or being replaced also depends on the person. The more you earn and the better educated you are, the safer your job is. It is important that society recognises these changes and adapts. You have only to look at Detroit, one of the richest places in 1950s, and now a place of poverty, they invested in structures not people. Contrast it with New York that invested in its financial sector.
History and common sense tells us that demand for high skilled jobs will continue to increase. The educational system should invest in the skills which we already know will be needed, such as computer programming and statistical analysis, but also in transferable skills that can be used in different sectors and in ways that we cannot yet predict. Debate over what should be taught and 'pushed' by the universities is necessary if we are to embrace the opportunities that the technology is opening up for us.
View a video clip of the lecture: http://www.futuretech.ox.ac.uk/replaced-robots
For more information about the Oxford Martin Programme on the Impacts of Future Technology, please visit http://www.futuretech.ox.ac.uk/welcome