Artificial Intelligence (AI) scientist Dr Anupam Guha, in a recent talk about AI in the Indian context, pointed out the positives and listed some of the perils of AI. He also said what we need is to lay out the framework to tackle potential challenges posed by AI.
“The thing in computer science is, the difference between what is very easy and what is impossible is not very intuitive, and hence it becomes a problem for policymakers to make policies around it, and here we need to be very cautious.”
To start with he said that AI is relevant to a lot of people, not necessarily only for people who are AI enthusiasts or researchers or scientists, and that’s what makes it so important. Another reason why it is important is that it’s a prototype-as-a-service science, and it’s not yet a well-researched field, and it can barely be called a discipline.
In terms of what exactly AI is, it is like a box, a signal goes in and the output is matched. A couple of billion signals are matched to their output, the box gets trained in a particular manner. While explaining the Conundrum of the Chinese Room, he said;
To an observer, it might look like the task requires intelligence or some kind of magic is happening in there, but what happens inside is actually just pattern matching.
What AI can do to jobs?
Speaking about what AI or automation or machine learning could do to jobs, he said that unlike 90’s, robots today can do a lot more tasks in a very sophisticated manner. And even though it will take more time than humans to perform a particular task, but it will be pay-less or wage-less. “No one will have to pay a robot”. One would only need to power it to run it, and with the algorithm replication, many more robots can perform the task in mass. And such cases are already happening.
He said rather, common jobs will go first. There have been studies in the US that several low skilled and high skilled jobs can be automated significantly leading to huge job losses. A study by Deloitte stated in 2016 that 35% of jobs in Britain are at high risk in the next two decades. Another 2016 McKinsey report said that the potential for automation in US at 75% for food services, 40% in services, 35% in education, and 30% in administration. “And US and UK are the developed nations, we are not even talking about Asian economies.”
Guha sited an example of Uber’s driverless cars, there was a notion that drivers will still be needed, for assistance and other things, and AI would make their jobs only easier. But, there are studies which estimate that about 6 lakh US drivers would lose jobs due driverless cars.
We have a psychological confidence that widespread of automation is going to spur job growth, and new technology is going to generate new jobs. “But this is simply not backed by data or any historical trends.”
It’s not just sectors like logistics and transport. Even professions like surgery, law, etc. would be affected. And, first to be hit will be the juniors and assistants. Speaking of India, Guha mentioned hiring in IT sector, which decreased by more than 40% in the last year with even sharper cuts predicted. However, he said that this has nothing to do with automation and the sector was bloated. Also, a very small chunk of employment. For India, he said that we should worry about farmers, who are half of the workforce in India. And with automation, “among the first to be hit, would be the farmers.”
That’s already happening, there are automated machines in India, which need no hands, and those machines can give seeds, soil and fertilizers. And, later reap the produce as well. “This is already affecting farm occupation”. Note that only 10% farmers in India own land, others actually work for wages.
What should we worry about?
“We need to think, what would happen when AI will be scaled up in agriculture, when automation will eventually become cheaper than human labour,” said Guha. “Is there a national policy of labour?”
He said that we should also worry about technology solutionism, and should understand that without addressing underlying issues, we will not be able to solve any problem.
Giving an example to think about technology solutionism, he said imagine that you were living in the 17th century. It was a bad time to live in, women did not have voting rights among other things. If a technology solution provider landed in this scenario and if people tried to explain the situation to them, what would they say? The right language to describe women’s problems haven’t evolved yet.
They might just end up saying, “women work too much; men work too little. Women are loaded with work”. And, the technology innovators would create a pressure cooker to solve that problem. That’s what technology solutionism does.
Similarly, we don’t know what the future of AI will be. We haven’t even developed the vocabulary to describe what the automated world would look like.
The way forward?
Guha said if AI leads to job losses, what could be the solutions. Is Universal Basic Income a solution? It might not be, because with UBI, a small set of people will still continue to make decisions about how much is equitable for all. What else? He said we can learn lessons from may be, the great depression in US, where we saw that both property and labour emerged out of that; or Verghese Kurien’s operation flood; or we can also learn from the company Mondragon cooperative, where all workers are owners of the company. The thing is, “we need to start being inventive when we think about AI,” said Guha.
We all don’t know what future will be like, we need a radical socio-economic change in AI, which has a potential to kill a significant percentage of Indian jobs and create a precarious situation for the hundreds of millions of India’s workers, both formal and informal, from farmers to engineers. We need politicians and policymakers to talk about this. He concluded, “if technologies like automation can be backed with political imagination and wisdom, it can lead to an emancipation of Indian labour.”