Computers powered by artificial intelligence are smart enough to threaten a range of jobs, whether computers developing treatment plans for cancer patients or Amazon.com opening a grocery store without checkout lines.
That impact is already being felt. The World Economic Forum expects automation, including AI, to result in the loss of at least 5 million jobs globally by 2020. In the view of Genpact (G) CEO NV “Tiger” Tyagarajan, however, the bigger question is how many jobs such technology will ultimately create.
Genpact, formerly a unit of General Electric (GE), provides what it calls “digitally powered” business process services, which relies on ever-more intelligent tech, to a range of corporate clients. He spoke with CBS MoneyWatch about how AI is affecting the economy.
CBS MoneyWatch: How smart are computers?
NV “Tiger” Tyagarajan: They have been smart for quite some time.You can leverage and use their smartness only if they have access to data that makes them smart. Only now are the computers and the software available that are capable of dealing with the masses of data.
So say you teach a computer that two times two equals four. How can you leverage that information?
Tyagarajan: Let’s take your example and let’s say the data show that when something comes from Boston, it’s two times two equals 4 and when something originates from New York it’s not 4, it’s 3.95, and when it originates at 2 p.m. it’s 3.85. So the reality is that the world is filled with data which has patterns.
Over time the computer says,“Huh, here’s a pattern that I recognize. Every time this and this and this takes place, this human being makes this decision The next time this happens, I am going to make the same decision.”
Are you saying AI can tell you the “what” but not the “why”?
Tyagarajan: You hit the nail on the head. The machine can look at the past and then say that’s how you always took decisions, so I am going to take the same decision. It can’t change that on the fly and say, “Sorry, I am going to make a different decision.” It can’t inject emotion and all other human characteristics. It cannot explain why it took the decision that it took.
Of course, not matter how “smart” technology is, you can’t compensate for incomplete or poor quality data. During the recent presidential election, for instance, polling data showed that Hillary Clinton, not Donald Trump, would likely win. Was that a failure of data or a failure of the algorithm that resulted in the erroneous conclusions that were drawn?
Tyagarajan: My perspective is that the data analysis showed what the data showed. If you did not go and ask the people who ultimately turned out that day to vote, then whatever your data shows has a mistake.
You’re saying that the conclusions or inferences computers make are only as good as the data they collect.
Tyagarajan: That’s one of the dangers. The whole world is excited about artificial intelligence. Many technologists are excited. Businesspeople are excited. I am excited.
However, it’s important to understand “Garbage In. Garbage Out.“ Data gets created by people. If that is bad, then actually it’s even more dangerous to have a machine make a decision rather than to have a human make a decision. Unless you tell a machine to check for garbage, it doesn’t know how to check for garbage.
Can AI learn from its mistakes?
Tyagarajan: Yes, it can learn, but a human being has to tell the computer, “Hey, you made a mistake, and here is the correction.” Then that goes back into the algorithm, and the algorithm then says “OK, I got it and the next time I will take that into account.” So over time mistakes are corrected.
What sorts of workers are likely to be replaced by machines, and will new jobs be created as a result?
Tyagarajan: The way that I would think about something like machine learning, computers, automation and so is that it’s going to change the nature of jobs. It’s going to create new jobs that we don’t even know exist.
So the good news is I think there will be enough jobs in the future. The bad news is that the new jobs will require new skills. Of that, there is no question.