Artificial Intelligence (AI) seems to be powering most of our modern day experiences. From playlist suggestions to voice-activated assistants, we are engulfed by the growing use of AI. Yet, the convenience is getting shrouded by the concerns about the fast-changing world around us. AI has invoked fears that there is an irreversible march of the machines. A corollary to the argument is also a worry about job displacements.
Dr Kenneth Kwok of A*STAR assures us that the fears of a complete takeover are unwarranted. To study the impact of the vast changes that AI will bring along, A*STAR has established the first of its kind initiative called the Human-Centric AI Programme under the stewardship of Dr Kwok.
“There is a lot of discussion about the possibilities and impact of AI. This is a result of some of the misconceptions about the reach and power of the technology to permanently change our lives,” Dr Kwok says. “We have to know a few important facts about AI to understand the implications better.” So, to help understand AI better, here are five (5) interesting things about AI that you probably did not know:
1. AI is older than you think
It might seem that AI is a recent advancement as smart technologies have only appeared recently. However, AI precedes probably anything you can imagine. “The field of AI was formally coined in a 1956 conference in Dartmouth College in New Hampshire. At the time, those attending the conference were really ambitious and were hoping to create ‘human-like intelligent machines’ within a summer!” shares Dr Kwok.
More than 60 years later we are still only scratching the surface. “If you consider the ultimate goal of AI is to reproduce human intelligent communications, then we have only achieved 10 to 15% of that objective,” Dr Kwok reveals. This is because early estimates had seriously underestimated the difficulty of creating a fully intelligent system. This was primarily due to the computational limitations of their time. Today, we have access to more data and computing power and hence the expectation is that things will progress faster. “Yet, with all the progress, we have just about assembled the building blocks,” Dr Kwok says.
2. Creating smart AI is hard
While we tend to ascribe a lot of credit to AI, the technology is still limited and preliminary. This is because human intelligence is a marvellously complicated feat. “Machines may have become capable of solving mathematical problems or playing strategy games like Chess, yet they are still far from semantic deep human understanding,” explains Dr Kwok.
Currently, AI is based on ‘Narrow AI Functionality’ where machines are as smart as the code that drives them. So, essentially the machines can only perform the narrow functions defined in the lines of code or programme running the functions.
One of the challenges of AI research is to produce human intelligence and empathy. Human intelligence stems from our ability to draw upon a wealth of knowledge and experiences that we store subconsciously. This allows us to understand the abstract and also make inferences about the context, things which we take for granted. As scientists have realised, is much harder to programme a computer to do so.
3. Singapore is the first country currently focusing on human-centric AI
Scientists all over the world may be focusing on different aspects of AI. However, researchers in Singapore are trying to look at the holistic impact and implications of AI.
What makes Singapore’s research unique is that it is human-centric. The goal is to produce AI that can demonstrate self-learning. “The real value of AI will be when we are able to assist people in a human-like intelligent way. For instance, anticipate the needs of an elderly and then provide an intelligent, human-like empathetic response,” explains Dr Kwok. This is the objective of the recently launched Human-Centric AI Programme at A*STAR.
“Our programme is unique as we are bringing all the aspects of AI together to come up with a human-centric approach,” Dr Kwok shares. Eventually, this programme will allow seamless human-computer interactions leading to intelligent AI companions or assistants to provide contextualised needs, much like a human companion.
4. People need to get used to an AI dominated job industry
As AI becomes pervasive, there will most definitely be a redistribution of jobs. “Some jobs may be redundant, but many more will be created,” assures Dr Kwok.
Adapting to such an environment is crucial as there is a predicted shift towards a knowledge-based economy. Even though some jobs will decline, there will be an increased demand for people with technical and engineering skills. “We will need knowledge engineers who will ‘teach’ the machines by creating knowledge banks,” Dr Kwok says about new job fields that will open up.
Jobs requiring human interaction such as nurses, teachers and caregivers would not be replaced in the foreseeable future. In fact, such jobs could be much easier in the future due to the prevalence of technology.
5. More time to play
In fact, AI may be our friend rather than a foe in the decades to come. With AI we may be able to work efficiently and even shorter work hours. “Before the industrial revolution, workers had to put in very long days at work with hard physical labour. Now, machines help relieve the burden. Similarly, in the future we may be able to enjoy more family or free time with the help of AI,” Dr Kwok opines.
In such a scenario, the only thing that would hold us back would be our own ability to be comfortable with technology.
Dr Kenneth Kwok
Dr Kenneth Kwok is Principal Scientist at the Institute of High Performance Computing (IHPC) at the Agency for Science, Technology and Research (A*STAR) and Programme Manager of the A*STAR Artificial Intelligence Programme (A*AI). He heads the Cognitive Systems group within the Social and Cognitive Computing department in IHPC, and is the PI of the Human-Centric AI (CHEEM) Programme under A*AI. Dr Kwok’s research interests are in cognitive computing – drawing inspiration from the cognitive sciences to build intelligent machines. He obtained his PhD in Psychology from Carnegie Mellon University working with James L McClelland. He also holds an MSc in Computing from Imperial College, University of London and his first degree was in Mathematics and Physics from King’s College, University of London.