Professor Domingos notes artificial intelligence needn't be feared and new jobs will replace old ones.

29 August 2018

 Artificial Intelligence (AI) is the subject of both fear and awe. No other technology trend has capture popular imagination the way AI has with proponents labelling it as a panacea for all problems and detractors, among them famous names such as Elon Musk, warning of the grave danger that AI poses to human society.

So the popular question is: “What is AI – human society’s saviour or the harbinger of doom, causing massive unemployment as machines take over human jobs?”

Pedro Domingos, Professor of Computer Science and Engineering at the University of Washington, feels that the question itself is framed in a wrong way.

“We tend to frame this as man vs machine. The real question is man with machine vs man without machine. If you have a horse you don’t ask how much faster than me it will run, rather you ask how far and how fast I can go if I ride the horse. AI is the horse.”

Professor Domingos is one of the world’s leading experts on machine learning, artificial intelligence, and big data and is the author of the hugely popular book on AI, The Master Algorithm. He is also a winner of the SIGKDD Innovation Award—the highest honour in data science—and a Fellow of the Association for the Advancement of Artificial Intelligence.

Taking part in the StarHub Speaker Series on July 17, the Professor delivered a master class on AI, explaining the principles that he has expounded in his book The Master Algorithm. After that he took part in a very informative conversation with Professor Annie Koh, Vice President, Office of Business Development, Singapore Management University, covering a wide gamut of issues relating to AI.

Discussing the important topic of job losses due to AI and automation with Professor Koh, Professor Domingos acknowledges that at the moment there is a lot of fear that AI will cause massive unemployment.

I think it’s important to keep a cool head. People are worried that AI is the continuation of automation – in fact the worst form of automation, he notes.

“There has been a lot of automation happening in economic activity over the past 200 years and there has been no lessening of employment.”

“Think about 200 years ago, almost everybody in the world was a farmer and agriculture was the most common job. Today, the number of jobs in agriculture is, maybe, one per cent of the total number of jobs that existed 200 years ago.” 

“Does that mean 99 per cent of us are unemployed? Not at all, the 99 per cent are employed in jobs that you couldn’t imagine 200 years ago like, for example, the job of a mobile app developer.”

So, he adds, it’s always easy to mention the jobs that are going to disappear rather than the new jobs that are going to appear because most of the times we don’t know what kind of new jobs a that technology breakthrough will bring forth.



Everyone talks about the disruptive power of AI, but what exactly does it disrupt?

Professor Domingos explains that what AI does is, in effect; lower the cost of (acquiring) intelligence which, in turn, vastly increases the number of ways in which this intelligence can be applied.

“It’s one thing, for example, to say that doctors will lose their jobs. But what is actually happening is that instead of going to the doctor once a year we may have a doctor on call 24x7 on our smart watches constantly advising us about our health. In the same way the Amazon virtual assistant doesn’t displace shopkeepers, instead it allows everybody to have an exclusive shopping experience,” he adds.

Professor Domingos, however, notes that a lot of traditional jobs will go. For example, self-driving cars will replace human drivers, including truck drivers. “There are millions of truck drivers and we are not going to be able teach all of them all to code and become IT experts. But then we don’t have to do that.”

“What will happen with self-driving trucks is that transportation cost will go down dramatically due to automation and prices of goods will fall. The knockdown effect of cheaper goods is that people will have more money in their hands so, maybe, they will eat out more or will have better houses. This will push up demand and there will be more jobs in the construction, food and beverage and other industries. Also, there will be completely new kinds of jobs coming up,” he adds.

The single most important thing to remember in this AI revolution, according to the Professor, is that apart from new jobs being created, the way we do our existing jobs will also be transformed. “The future belongs to those who know their job very well, are good in their area of speciality and, at the same time, know how to apply AI to their jobs. The way you do the job is you automate yourself.” 

Strategy Required

Professor Domingos agrees with Professor Koh’s observation that with AI having such a disruptive impact, every company and country needs to have an AI strategy in place.

The best way forward is to have a two-pronged approach, he adds. One is to try to use AI for every business function; some of them might work and some might not, it will be a journey of discovery. Theoretically any function where there is an input and an output, can potentially use machine learning, he notes.

“The other aspect is how you make AI a part of your external strategy. AI will be taken for granted,” he adds.

Professor Domingos notes that the big IT companies, Google, Facebook and Amazon and others have the most data today and good data is a pre-requisite for good AI. “Every company thinks they know their customers well and in some respects they do. But these mega IT companies know your customers in many ways that you don’t from their own interactions with the customers.

“They have the algorithms and they have the data and there are only a few things an independent company can do to challenge them,” he adds.

So how does a small and medium-sized enterprise (SME) go forward with its AI strategy?

One way is to dogo it alone without partnering any of these big tech companies. “This strategy is actually dangerous. Unless you are in a sector where your core knowledge is more important than the broad knowledge possessed by these big companies, it would be difficult to survive.”

“Another option and my expectation is that most companies will take this route, which is to align themselves with these large companies and lease some of their data,  and use it on a platform and also use their virtual assistant.

“This is could work but the worry is that these tech companies may capture most of the value because it is their data.”

He adds that the third option, which is likely to be the most risky one, is that smaller companies can make alliances with other companies that have similar interests so as to form a large eco-system that can compete with the likes of Alibaba, Google, Facebook and Amazon.

“This is of course hard to do as it takes a lot of technology and a lot of data, but if you succeed you potentially might end up as a strong entity.”


No Doomsday

Responding to a question from the audience, Professor Domingos categorically ruled out the doomsday scenario of machines exterminating humans. “The machines are achieving goals that we set for them, they are automation and an extension of us, our brains and them turning evil is highly unlikely,” he adds.

However what worries the Professor is the damage AI can cause in the hands of evil people.

“Evil people with access to AI are a much bigger worry than AI on its own. Criminals will use AI for their purposes. A corrupt regime will use AI for its survival.” Professor Domingos adds that in cyber security hackers have started using AI programmes to steal data and on the other side machine learning is being used by security experts to detect and deter such attacks.

He notes that security of data is hugely important and companies need to realise that if they cannot keep their data secure nobody will trust them with their data.

To a question by Professor Koh on what kind of strategy Singapore can take with regards to AI, Professor Domingos says that while every country has its own strategy, China has been “extraordinarily aggressive in AI”.

“Ten years ago China was nowhere in the picture in AI. Today, in some respects, it is starting to close the gap with the US in AI research and deployment. Eric Schmidt, former executive chairman of Google, has said that in another 10 years’ time China will be in the forefront of AI,” he adds.

Discussing why China is racing ahead, the Professor notes that in order to have good AI there is a need for large data sets and China has a lot of data. “In China there is much closer cooperation between top companies and the government in AI, while in the US it’s just the opposite with the companies and government not trusting each other.”

So in this situation, if you are not a China or the US how do you compete? Professor Domingos feels the way to do that is to forge alliances with similar sized and similar minded countries. “I think one huge advantage that smaller countries (like Singapore) have is that they are more nimble. Nimbleness is extraordinarily important and that is something the bigger countries do not have. “If you can combine AI with deep knowledge and domain expertise you can beat everyone else,” he adds.

One of the key takeaways from the interesting conversation between the two professors is that the biggest difference between AI and humans is that the former lacks common sense. “People worry that computers will get smart and take over the world but the real problem is that machines are stupid and make mistakes like how a self-driving car runs over someone because they don’t see them,” he adds. It is very difficult to instil common sense in machines.


Follow StarHub Business on Linkedin for the latest business updates.

Is your IT Infrastructure AI-ready?

Top 3 considerations in getting your organisation’s IT infrastructure AI-ready.

Read more
Futureproofing your business with data analytics & AI

Key insights from expert speakers at NTUC UFLX forum.

Read more
Healthcare is rapidly transforming with AI and IoT

How can a finite number of hospitals and professionals scale up with AI and IoT.

Read more