'We'll see societal changes when A.I. affects people's lives in really meaningful ways'

Mohammad Rahman, Associate Professor of Management, Purdue Krannert School of Management spoke to Business Today's Rajeev Dubey on the evolution of Aritificial Intelligence. Edited excerpts.  

Are you on the Ayes or the Nays side of AI debate? Would you concur with Zuckerberg or with Elon Musk?
It's an interesting question. My thinking on this probably resonates with both of them. I can't just say I can agree with one of them. To be very frank, I don't know their full thinking. But, let me explain my position on this. That might be easier. So, I don't think we can just sit here and not worry about thinking through the implications and the policies we need with the development of AI and how it is going to transform our society, our workforce, and our mobility. Fifty to 100 years later, a lot of this may be second nature, but we've got to get through that 50 years. So, we need to think through what it means. At the same time, I don't think we should say, 'OK, timeout we don't let this technology work.' I don't think that is what Elon Musk is saying, that we just stop. And, I don't think Zuckerberg thinks that we don't need any sort of regulations in the thinking through this process.

If you think about today, the Tesla, Elon Musk's car, is able to detect crash probabilities and stop the car. Now, none of us want to get involved in a car accident. It's a black and white decision. We can train AI how to do this; avoiding getting in car accidents. We should. Now, when a car accident happens, how often do people passing by stop or get out to help those affected? It's very, very judgmental. My response may be different depending on the day, and my priorities and where I need to be. How many times do you see someone needing help on the road, who maybe needs help with a flat tire? How many times do you stop? These are judgmental things, and these are simple judgmental things, but there are lot more complex ones that need to be addressed with AI related possibilities. If we don't think through the revolutions of choices and judgements that we have as humans, and somehow that is not part of the design, that is going to be a problem. If you think we are all going to buy one color iPhone, that is how you can potentially think, but that is not true. There are potentially three or four colors and everyone is buying one. It is not the same thing. So judgment is very important.

What's the state of evolution of AI right now?
To me, AI is when you are able to let the machine make certain decisions that are not so codified. For example, in a crash detection system, it is calculating the probability of a crash and making a decision to stop the car. That is AI. Where is the state of evolution? I think we are still in the world of small worlds with AI. As human beings who deal with a lot of big-world decisions, AI is not ready to step in. AI is specific to the activity such as the crash detection system. That crash detection system is really good at figuring what cars are following each other and what is going to happen when the car in front of you stops. That is precisely what AI is trained for. Learning itself without any kind of training data, that is not out there. We need a lot of data to train AI so that it can deal with situations as they come up based on what it has seen before. That is where we stand today.

In your view, which of the world's primary AI tool providers--IBM, Google, Microsoft, AWS has a headstart vis a vis others? What do you see as the SWOT of each of these?
This is an opinion question and I'm not privy to everything going on inside these companies. But from what we see on the outside, I think IBM is a little ahead. I think IBM has invested a lot and took a risk up front. On the other side, the AI requires a lot of learning from queues and contextual data. IBM is relying on a lot of these from its partners, such as those in health care and with GM on the OnStar system to do shopping and planning. Depending on how partners and how partnerships come about, it could be changing their competitive advantage. In my view, Google has a lot more contextual information about people than any of these companies -  from Gmail, to searches to trying to get into the transaction side of the business - and they've partnered with Walmart to order things from Walmart. Google Home now knows what temperature you have inside your house. As that evolves, companies will take a look at how much data do I have, and how will I train that system. There is a big competitive advantage in having complexity in the data that you have. So, I think Google may be able to catch up. I think Amazon is heading that way, too. Alexa is pushed everywhere. Microsoft is a very good software company, and at this moment, it seems they are kind of like IBM where they have to rely on other companies because they don't have a lot of contextual data process. These companies are all going to fight for leadership in AI. And as consumers, we are probably not going to stick to just one of them.

Today, a lot of chatbots, data analytics, big data, etc. are being passed off as AI. What's your view?
In my view a chatbot, where you have a structured question and a structured answer, is not AI. Analytics and big data are going to be how we see the world and make decisions, but just having data analytics is not having AI. AI to me is where you have a system that is able to make decisions when facing a decision-making scenario.

When do you expect the real AI revolution to begin?
I think everything has a silent phase, and then a buzz phase, and then an application phase. Machine learning and artificial intelligence was really picking up in the '80s, and it sort of died down, and it is now slowly coming back. I think we are going to see societal changes, or you can call it an evolution, when it affects people's lives in really meaningful ways. That could start with self-driving cars. And some factories are already automated, or warehouses like Amazon, where you have automated robots doing stuff. As that becomes more prevalent and forces us to move away from tasks that people have done for many years, that is when the revolution starts.

Startups seems to be taking to AI faster than established companies. Any reason?
That is natural, because in established companies you always have to worry about priorities that are going on right now. I have to generate income. I have to make my shareholders happy. Places like Google allow workers to have pet projects, but the pet projects are not there for everyone for five days a week. The company would be in trouble. It is natural to see a lot of start-ups. In a start-up culture, you basically let a thousand flowers bloom, but we don't see all thousand when we are done. We see a few that are surviving. To me this is a such an unstructured problem, something that we are trying to push on to the next horizon. It ought to be individuals and small companies really trying to do it right.

Do you have any views on the adoption of AI in India?
India is a fairly advanced country when it comes to technology because of outsourcing and the workforce. Certainly, you will see a lot of push there as well. I would not be surprised if many things are rapidly deployed to India. But, like with any other technology there are contextual things that are different. AI is kind of different. What we have in the U.S. works well for us in the U.S. but might not work well in China. That's the lesson Uber learned in China. There are cultural things. In India, there are some bright spots for culture to take off, but there are also many people who are trying survive by doing basic agriculture. In a country where you have a large population, where not everyone is skilled and employed to their fullest extent to where they are productive, but then you get AI; it will displace people who are kind of well off right now in the relative sense of the society. I think it can create a lot of chaos in society. Being one of the largest democracies, I think that makes it complicated as well.

What's your view of use of AI in China and specifically China's approach to it. How different is it?
It seems AI is identified as data tracking and data generation that is not necessarily AI. I do know that in China that there are a lot of factories where they are replacing humans with AI robots. Especially repetitive tasks. Robots are good at repetitive tasks, but dexterity is something it's not very good at for instance. My take on China, because of regulation, because of society, because of the government strategy, things take off faster there. If the government thinks they are going to do something, it is going to happen quickly. Where in a democracy there would be a lot of push back or discussions. There are both sides to it. I'm sure there will be some advancements, and there is a lot of money in the economy in China.

Another point, is that warfare is based on a lot of AI driven systems. The next generation of influence in the world and who is the mega power is really dependent on how much innovation around cyber war or AI war, such as unmanned planes bombing places and being able to detect places. China is very well positioned to take on things because of resources.