SINGAPORE: Crafted to resemble human-likeness with natural looking skin and hair, Nadine, described by its makers as a social humanoid, greets a person when he enters the room, makes eye contact and remembers his face.
She smiles and listens intently as one speaks, cracking a joke occasionally. But when provoked, she gets upset too.
Made in Singapore, the artificial intelligence robot, which was unveiled by the Nanyang Technological University's (NTU) Institute for Media Innovation in 2015, has made significant strides in the past three years, including the ability to recognise faces and remember conversations.
Plans are underway for another public showcase next month, and researchers hope that before long, Nadine will be able to walk as well as identify and grasp objects with her hands.
Once the domain of science fiction books and movies, the AI revolution has well and truly arrived — and it is not just in the form of walking and talking robots. The rise of AI has pervaded areas ranging from healthcare, retail, transport and banking to food and beverage as well as dating services, to name a few.
By 2020, AI is projected to create 2.3 million new jobs worldwide while eliminating 1.8 million traditional jobs, according to research firm Gartner. This was a point echoed by a recent study by the World Economic Forum which said the development of AI technologies will disrupt jobs but at the same time usher in unprecedented new opportunities.
Around the world including in Singapore, research on AI as well as its application can be seen in almost every sector.
In healthcare for instance, scientists at A*star's Genome Institute of Singapore have discovered the use of AI to help pinpoint roots of gastric cancer.
The project, the findings of which were published last month, involved the creation of two new AI methods to scan the entire genomes of 212 gastric cancer tumours in a few months. According to the researchers, such an analysis would have taken 30 years to complete on a standard modern computer.
Chinese bike sharing platform Mobike, which has operations in Singapore, uses an AI data monitoring platform to make forecasts of supply and demand for bike rentals, and help improve the firm's operational efficiency.
Many developers have also turned to AI to create apps catering to anything from consumers' social and emotional needs to gastronomical desires.
For example, Replika uses AI to create a chatbot in a person's likeness — learning from users to match their personality and becoming their best friend, as the app's developers put it.
Homegrown dating service provider Lunch Actually has also launched Viola.AI, which analyses a couple's compatibility and dishes out relationship advice, among other things.
Another Singapore startup Tabsquare recently unveiled "Aiden", a "first of its kind AI engine" which enhances a guest's dining experiences by constantly learning and building taste profiles and preferences.
According to research firm IDC, worldwide spending on cognitive and AI systems will grow to about US$19.1 billion this year, a 54 per cent jump compared with the amount last year. By 2021, the spending is projected to increase exponentially to reach US$52.2 billion.
THE BEGINNINGS OF AI
For decades, the concept of AI has been around in literature and popular films. Even before the technology was available, people have been seized by the idea and its endless possibilities.
The roots of AI research can be traced back to the 1950s, said Professor Ong Yew Soon, chair of NTU's School of Computer Science and Engineering.
At that time, computer scientist and mathematician Alan Turing proposed a test to measure a machine's ability to exhibit intelligent behaviour that is indistinguishable from that of a human.
The first working AI programmes — which were able to play checkers and chess — were written in 1951. However, funding dried up eventually, as the lack of technology and slow computing powers held back the development of AI, said Prof Ong. Progress continued to be tepid in the following decades.
Professor Kevyn Yong, dean of ESSEC Business School Asia Pacific, added:
It's quite clear that the AI scientists (then) were limited by computational power and data – both of which are critical to advancing AI technologies.
The late 1990s saw the resurgence of AI, thanks to the advent of the Internet and advances in computational power among other factors, experts said.
In 1997, IBM's Deep Blue super computer beat then-world chess champion Garry Kasparov. The feat made international headlines, as it marked the first time AI defeated the best human chess player.
The convergence of several factors in recent years has led to greater momentum. "Computers became faster, algorithms improved, and a large amount of data became increasingly accessible. New advances followed, with deep learning methods starting to dominate around 2012," said Prof Ong.
Dr Justin Chan and Mr John DeCleene at research company Data Driven Investor and OC Horizon noted the slew of AI products which were rolled out in the early part of this decade — Google's first self-driving car, Microsoft's Kinect for Xbox 360, and Apple's Siri.
Last year, the development of AI achieved another milestone when AlphaGo — a computer programme that plays the complicated strategy game Go — won a three-game match with world number one player Ke Jie.
The rapid rise of AI in the last few years has resulted in such technologies being infused in everyday life, experts said.
Prof Ong noted that most of the current AI technologies focus on solving specific tasks where AI can surpass humans. These range from photo-tagging on social media, chatbots, loan approvals to language translation and digital assistants, he said.
Ms Sharala Axryd, chief executive of ASEAN Data Analytics Exchange, added: "(AI) tools that we use every day that perhaps we take for granted (include) Siri and (Amazon's virtual assistant) Alexa.
The widely-used Google maps app also uses AI, and it has helped to improve people's daily commutes, noted Prof Yong.
Across industries, companies are leveraging AI to enhance their business by performing repetitive tasks or improving efficiencies, among other things. Some firms and governments around the world are also turning to AI to forecast trends — a service currently provided by homegrown AI tech startup Scry, for example.
In the transport sector, Mobike's Magic Cube, for instance, manages the company's fleet of 9 million bicycles around the world. The firm provides up to 30 million rides a day in over 200 cities, generating nearly 40 terabytes of data daily.
The AI technology enables the company to pinpoint optimal locations and times to place its smart bikes, as well as map out the most efficient routes for the operations team to move around each city to manage the fleet, Mobike said. It added that the data can also be shared with city planners to assist them in planning transport infrastructure.
Cybersecurity and banking are other areas where AI is making its mark.
Multinational firm Darktrace — which has its regional headquarters in Singapore — created the first AI technology to detect cyber threats. The technology is able to learn and pick up abnormal activity in a network, without the need for prior knowledge of the threats.
Homegrown AI tech startup Sense Infosys, for example, also uses its AI system for streamlining data and identifying fraud risk anomalies, among other things.
Meanwhile, OCBC Bank has established an AI laboratory this year to look into applying the technologies to banking services.
Emma, a chatbot for renovation and home loans, is already up and running. Launched at the start of last year, it has since answered more than 80,000 enquires and helped to secure more than S$100 million in home loans, said Mr Ken Wong, OCBC Bank's Head of AI Lab (Fintech and Innovation Group).
MYTHS AROUND AI
The rapid development of AI worldwide and its pervasiveness in everyday life has created some anxiety among the public, no thanks to the doomsday scenario often portrayed in books or Hollywood movies where AI robots run amok for instance.
"I think we're still very far away from 'sentient' robots that we often see in Sci-Fi movies," said Prof Yong.
To create such machines, scientists must first solve the mysteries of the human brain and how intelligence works in every social and cultural context imaginable. Then, they will need to build the technology, including the learning algorithms, based on the "perfect training dataset", Prof Yong pointed out. "(These are) theoretically-possible … (but) I can't see scientific feasibility anytime soon," he added.
At this point, AI is extremely capable of analysing available data and recognising pattern, but it cannot make inferences or judgements, Prof Yong reiterated.
NTU's Prof Ong said that while AI can perform particular tasks like humans such as cooking specific meals, it is very unlikely that machines will be able to exhibit "broadly applicable intelligence comparable to or exceeding that of humans" — not for the next 20 years at least, he added.
"Currently, we are seeing a rapid progress in the field of specialised AI … which will most likely continue in the years to come," he said, adding:
In other words, we may not expect a robot replacing a chef in the very near future. Nevertheless, it is to be expected that machines will reach and exceed human performance on more and more tasks.
Presently, AI technology has not reached a point where it is able to develop new output possibilities beyond what it is programmed to do, said Dr Chan and Mr DeCleene. Even for the area of self-driving cars, where there is "significant hype", introducing a high-quality AI product requires a significant amount of testing and further development, they said.
Although AI is advancing at a breakneck speed, being human also means being emotional — to be able to hate, love, be angry or sad, said Professor Louis Phee.
Prof Phee, who is the dean of NTU's College of Engineering, had invented and commercialised the world's first scarless surgery robot for stomach tumours. "These traits are difficult to understand, even for humans. They would be the challenges for AI scientists to solve in the future," he added.
Prof Nadia Thalman, who created Nadine the social humanoid, noted that AI cannot surpass human intelligence "as long as the computers are made with processor chips".
"I do not think it can be like a human brain, because it is all fast logic calculations, comparisons, statistics and programming, programmed by humans," said Prof Thalman, who is also the director of NTU's Institute for Media Innovation.
But she believes the gap between AI and humans could narrow considerably "if biotechnology improves over time".
"We can imagine having robots, for example, built with real cells and organs … then if robots are bio-robots, the difference between humans and robots capacity will be lesser and lesser," she said. But she stressed:
The myth that a robot feels or have desires — all are lies. A machine feels nothing and has no awareness of its condition in life.
IMPACT ON JOBS
While worries over AI taking over the world from humans are far-fetched, there are legitimate concerns about its impact on jobs — specifically, taking over roles which used to be performed by humans.
In a research note, Dr Chan and Mr DeCleene predicted that roles in areas such as telemarketing, customer support, bookkeeping, reception work, sales and market research among others are "most likely" to be replaced by AI.
These roles have common underlying traits: They are "highly susceptible to automation", the tasks involved are repetitive and there is a "high focus on manual efforts compared to thought process".
Jobs that are least likely to be replaced by AI include managerial roles, positions that require expertise, and those that involve "unpredictable physical work".
Dr Lin Hsuan-Tien, chief data scientist at AI startup Appier, noted that job displacements and the creation of new roles are inevitable in the face of new technologies.
For example, the role of a data scientist did not exist 15 years ago. Today, however, there is a shortage of such professionals in the market, he said.
Even as some jobs are destroyed by the rise of AI, new ones will be created. For a start, there will be demand for people to manage data and make sense of it, he said.
Mr Wong, from OCBC Bank, pointed out how his bank has hired more people for AI-related jobs since last year:
Besides data scientists, who develop AI algorithms and models, we are also hiring more engineers to manage these AI applications, and business managers who understand how to leverage AI within the business to increase efficiencies and improve customer experiences,
Apart from massively changing the nature and type of jobs, the rise of AI will bring about profound changes to societies around the world, the experts said.
It can boost economic growth rates and increase productivity, for example. With its wide application in almost every aspect of people's lives, the possibilities are infinite.
Ms Axryd cited healthcare as an industry that would benefit substantially from AI. Patients can be diagnosed more quickly and accurately by apps designed to identify illnesses, she noted.
Meanwhile, businesses and organisations can also perform better when AI is used in areas that could be impeded by human biases, said Prof Yong.
"Information processing biases often hinder organisations from innovation because (they develop) biased customer insights due to human biases," he said.
AI technologies can improve how organisations collect and process customer data by eliminating human biases. As a result, they can gather insights that create "real value for customers and society at large", he added.
While there are untold benefits which AI could bring, there are also risks. For example, individuals could exploit AI for nefarious deeds such as hacking and weapons development, said Dr Chan and Mr DeCleene.
For now, the experts are divided on how fast the AI revolution will take place. Prof Ong noted that some experts predict "with 90 per cent confidence" that by 2070, AI can do most jobs at least as well as a typical human.
Dr Yu Chih-Han, the chief executive and co-founder of Appier AI, believes it could take even shorter, although it will be a "gradual and progressive revolution". He said:
In 10 years, we will see a very different world.
There is no "theoretical ceiling" on how far AI can progress, Prof Yong reiterated.
"But if I had to choose one, it would be when human civilisation reaches the perfect understanding of how human intelligence works – and in so doing, it should be theoretically possible for human civilisation to build the 'perfect' AI; one that is completely indistinguishable from humans in every way possible," he said.