Duncan Roberts: You have said that artificial intelligence is an “enabler of productivity” and that job losses in some areas are inevitable in the near future. So, what should youngsters be studying or learning if they want to guarantee employment?
Anand Rao: We are moving away from reading or learning a body of knowledge, whether its physics, chemistry, biology, to how do you learn about learning--how do you practically think about problems, breaking them down quickly to into different pieces to bring the right knowledge to bear. Those are some of the skills you will need. And also, how do you deal with people, how do you empathise with them. So, in addition to the content we are teaching, we also should be teaching and learning about new ways of solving problems using technology, using people and all sorts of combinations.
So, learn math at a very fundamental level; it is the underlying base. Then, I would say, philosophy, which teaches you how to think critically, to analyse and debate--I think that becomes more important. And third is human psychology and emotional intelligence, behaviour and how to change behaviour. Those would be the primary things that kids need to learn. Today, given all the search engines available, you are expected to know anything from anywhere. But the skill then lies in how you bring together, from all the billions of documents, what ones are the most relevant to the particular topic. It is a very different skill, and we believe it is going to accelerate even more in the next 15 or 20 years.
DR: Artificial intelligence is already being commonly used in a variety of applications--you have cited Google mail and Netflix algorithms that suggest movies to users. But are consumers fully aware that AI is being used?
AR: I think most people are not aware of what AI actually is. In the common press I would say there is almost this Hollywood version of what AI is; where it is something like HAL or Robocop. Most people don’t realise that the software behind music recommendations, for example, uses AI. What’s still coming is the use of some of that AI in the enterprise world. There is at least a 3 to 5-year gap between the consumer world and the enterprise world. I think enterprises can get smarter in the use of data.
I think in the consumer space it will push even further. We all carry these so-called “smart phones”. But they’re not really smart. Most people probably only use half a dozen apps on a regular basis, and even that half-dozen are not aware of what the other apps do. That is one way that AI will fundamentally change the way we use smart phones, whether it's voice commands or more intelligent systems communicating between the apps. The device itself may disappear. Maybe you will just need an earphone, and no other physical device to communicate. Some guys like Elon Musk are thinking of neural implants, but that’s much further out.
DR: In your paper on “A Strategist’s Guide to Artificial Intelligence” you say that decision makers may face a change if models suggested to them by augmented intelligence do not “match their past experience or gut feelings”. Will future leaders require training, or a radical mind shift to help them be open to such alternatives?
AR: In the business world we make multi-billion-dollar decisions with our gut feeling, maybe with some data, but not correlated data. How do you get that kind of thing into more data-driven decision making? Well, in a flight simulator you crash multiple times, but eventually you learn and then you are able to fly in the physical world. What we’re trying to do for a few companies now is to create the same kind of simulation for executives using gamifying strategy. So, instead of reacting to what the competition is doing or what the government is doing, they can be much more pro-active in deciding what is required in a simulated environment.
We did it for a large auto manufacturer, which was looking to invest in autonomous vehicles. Initially we thought there would be a negative tendency towards autonomous vehicles. But doing all these simulations, people ended up saying, “hey, there are 23,000 deaths due to road accidents in the US. Of those, 92% are due to driver error. Now, if we can avoid even 50% of those 92%, that’s a huge amount.” So, you’re probably talking about saving 10 or 15,000 deaths a year using autonomous vehicles. Now, isn’t that a good thing for the government to be promoting? So, that’s the story now--it’s no longer primarily about cool technology and saving people time; it’s about saving lives. The question becomes, is it morally wrong for the government not to support an initiative that is going to save 10 or 15,000 lives a year. But we are at an early stage with this, like we were in 1984 with computers. In the future, I think we’ll see more and more businesses try this out.