The coming age of Robotics and Automation
Age of Automation
Marc Andreessen said in 2011, Software is eating the world. As the years go by, it is becoming more and more clear what that software-eaten world is going to look like.
The purpose of software is to automate. As software development practices advances, the kinds of automation that are being attempted are ever more complicated including things which we were used to thinking as exclusive domains of humans not so long ago. There are already a lot of specialized software that’s better than humans at specific tasks. The best chess player has been a computer for a while now and will probably always be. The best Go player is now a software as well. Air traffic control, defense systems, nuclear plants are all controlled by software to a large extent. Also, in the very near future, software will be driving cars on our roads as well.
However, these software are each an isolated piece. The chess playing software cannot yet do air traffic control. Applying learning from one area into another, which humans do effortlessly, is still an open problem in computer science. Such general intelligence which transfers from one skill to the another, is still a specialization of humans. However, software has been steadily gaining at that. There are several companies working on what is called a universal learning machine.
Research and development of such software is often covered under the broad terminology of Artificial Intelligence or AI. However, AI is just a fancy name for automation using software. Because of its name and assumed connotations it has been a source of disappointment to some.
Even within AI there are two schools of thought – one called Weak AI which is focused on a narrow task. All current achievements of AI can be categorized as Weak AI.
In contrast, the advocates of Strong AI believe that our minds themselves are sort of a computer. And that a good simulation of mind is not different from mind itself and should be considered sentient or conscious. Needless to say, we are nowhere near achieving a Strong AI. Some even argue that, hypothetically, whenever we achieve something like a Strong AI, we will very quickly have something called Superintelligence which will far surpass humans.
Coming back to the present time, even with a Weak AI, we are seeing an increasing replacement of humans by software. That raises an important social and economic question – what will humans, replaced by such automation, do? The problem will become even more serious if Strong AI or Superintelligence were to become a reality.
A majority of humans depend on their work to earn their livelihood. They sell their skills and time in the market to earn money which allows them and their families to earn the necessities and hopefully some luxuries of life. If large number of people cannot find work, that’s a direct threat to their survival.
Over the ages, increasing efficiency has allowed humans more free time away from the problem of earning a livelihood. That in turn led to the rise of art, music and culture. However, the current wave of technical progress is different. It is possible a large number of humans will be left completely redundant. That can become a huge social and economic problem.
The most commonly proposed solution to this problem so far has been paying every one a fixed income often called Universal Basic Income or UBI.
Alternatively, it could be paid out of fees charged for using shared assets like air, water, judicial system, financial system etc. and money thus collected equally divided among citizens.
UBI is also being seen as a key tool for poverty alleviation. Governments anyway pay subsidies to select industries or individuals. That could be made into a direct cash transfer.
UBI is seen by some as a response to a fundamental shift in our society where work no longer is the defining purpose of life.
However, UBI is far from a resolved issue. Yuval Noah Harari in a recent TED dialogue interview raised some very pertinent questions about the viability of the idea of UBI. He said, the question of what amount will be considered as a basic income is not really clear. Also, under the current political structure of nation states UBI cannot be universal – “will textile workers made redundant in Bangladesh because of American companies using robots be compensated by the American government?” UBI may end up creating a more unequal world than what we have currently. Those are some serious questions.
What should be automated?
All this raises the question of what should, we as a society, automate? Currently, that question is left to the technologists. Whatever is possible to automate by current technology gets automated. The bigger question facing us is: what things are actually worth automating?
Clearly, anything where chance of human error is a big risk, while software can reduce the error rate, should qualify. Things like scientific calculations, rocket launches, GPS and such. Also, tasks which are too dangerous for humans to do e.g. nuclear accident sites, sewage and industrial waste cleanup, some types of mining, etc. should certainly be targeted to be done by robots. In contrast, we should be reserved in automating things where human judgment or an emotional touch is required e.g. customer service, health care etc.
Perhaps, the tide of technology is too strong for us to try to stop or change its course. However, governments can promote automation in the right kind of areas using various incentives while discouraging it elsewhere.