Artificial intelligence is growing smarter by the second, and soon it will be able to outsmart us and take our jobs -- or at least many fear it will.

But this sentiment is largely premature, as AI in its current state is “not the beginnings of general intelligence,” Philip Evans, a senior adviser to the Boston Consulting Group, told The Korea Herald.

“There is a widespread perception that AI is converging on human intelligence, that there is a process of progress by which AI will become like human intelligence in the foreseeable future. I’m frankly quite skeptical of that,” said the strategy expert, who founded BCG’s media and internet sectors.

Rather than thinking about AI as a general substitute for human labor, Evans said, it is important for businesses to understand the capabilities of AI and figure out where those capabilities are better than those of humans.

“For example, in a bank, identifying a fraudulent credit card transaction is a classification problem. But it’s very difficult in practice, and it’s really important in credit cards to identify fraud in real time, because you’ve got to respond immediately and stop the fraud before it happens or before it spreads,” he said, pointing to classification as one of the skills that AI has been able to acquire.

“It turns out that banks are able to use AI and deep learning in order to monitor the pattern of transactions and to learn patterns associated with fraud, and to stop them much faster than humans could.”

On the other hand, AI still trails far behind humans in sensory motor control and abstraction, according to Evans. 

Without making this type of distinction, many businesses face the common pitfall of trying to get onto the AI bandwagon without a specific application in mind. 

“You really need to start from the business, not from the technology. It is fundamentally wrong to say, ‘How can we use AI?’ That results in silly applications,” Evans said.

The situation, however, is opposite for tech companies compared to other industries, as advancing AI technologies has become the yardstick of the firm’s very survival.

“When Apple, Amazon and Microsoft compete to put a device in the home that you talk to, they are really playing a highly strategic game and the aim is not just to make a better product. The aim is beat the competition, because they know one company will win.” 

As tech companies both in Korea and abroad continue to roll out sophisticated voice-activated assistants and other AI services that tie together various devices, electronics brands have no choice but to explore how they can improve their AI competitiveness. 

“The major Korean electronics companies are fighting a continuous battle to maintain product differentiation and avoid becoming a mere commodity. And AI, from their point of view, is a key way to do that. Therefore they are trying to put AI in as many products as they possibly can,” Evans said.

“If it works, and they don’t do it, they have a huge downside. If it doesn’t work, they’ve just added some unnecessary features, which is something that companies have been doing for decades. It doesn’t cost much.”

So what kind of skills will be necessary in the future to secure jobs that AI cannot replace? Evans says it is necessary to look at what AI will never be able to do.

“There are some things that machines are inherently incapable of doing and the most obvious one is emotion,” he said. “They don’t feel emotion, and we know that they’re faking, but we don’t care ... my own view is that it’s certainly great to be a data scientist, but it will be human emotional intelligence that will become more important in the future as the purely cognitive skills become less important.”

Evans also warned against the dangers of using AI to make ultimate decisions in areas such as human resource management because AI ultimately bases its decisions on decision-making patterns of the past. If past decisions were biased, AI systems will learn those patterns and lock in those biases in the future -- which is why for now, companies need to be wary about lending too many tasks to AI.

“I would not want to live in a world where the final decisions were being made by machines,” Evans said. 

By Won Ho-jung (hjwon@heraldcorp.com)