Making AI Work for you
Singapore, September 10, 2018
As technologies such as the steam engine, electricity and the World Wide Web swept across the globe,
they ushered in the first three industrial revolutions. With the ever-expanding capabilities of artificial intelligence (AI), there is growing consensus that the fourth industrial revolution will be powered by AI.
But just as steam engines took very different forms depending on whether they were used aboard boats, trains or on the factory floor, AI is not a single technology but an umbrella term that encompasses a broad range of techniques, from machine learning to computer vision and natural language processing (NLP).
Although AI will undoubtedly have an impact on all industries across the board, each industry will use different kinds of AI depending on its own needs, explained Dr Tan Kar Han, Head of Product Research and Development at NCS.
AI, tailor-made for the job
“The recent wave of advances in AI has really brought the capacity of computer vision and speech recognition systems to a different level of sophistication,” Dr Tan said. “Computer vision, in particular, is automating many tasks where you previously needed humans to look at a picture.”
These advances could mean that AI will soon be doing tasks that until now were only granted to the most highly educated and trustworthy humans in society. The practice of medicine, for example, could
be enhanced by bringing AI on board. In fact, researchers have developed a deep-learning algorithm that can outperform radiologists at diagnosing pneumonia from chest X-rays1.
“There are countries with vast areas where you don’t have big hospitals and doctors to cover every village. We might be able to employ a remote diagnosis station outfitted with AI and only escalate
cases that are critical,” added Dr Tan.
While computer vision may ease the workload for doctors and other fields that rely on humans to visually assess situations, AI in the form of NLP can transform sectors like customer service management. NLP is a subset of AI that enables machines to interpret human conversation and respond in an appropriate manner.
NLP will not only allow machines to answer commonly asked questions, but can also be used to assess emotions in text to help them to decide when to escalate a case to a human operator. The Gov.sg chatbot launched by the Singapore government, for example, allows the public to ask questions, file reports and contact civil servants through Facebook messenger2.
“These are the main things impacting us in the first wave of AI,” continued Dr Tan. “A larger wave is coming in robotics, related to self-driving cars, ships and airplanes.”
Keeping humans in the loop
Regardless of the AI technique used, humans aren’t likely to become obsolete in the workplace. They will, however, need to get used to working alongside their virtual colleagues, Dr Tan said.
“It’s not so much that humans will be displaced; rather, automation will be primarily used to amplify human labour. With automation, we will be able to cover a lot more ground that humans couldn’t before,” he added.
Nonetheless, any system built should consider cases where we would want humans to be able to override AI if it is not doing the right thing, Dr Tan said.
“I believe we want to have human supervision,” he explained. “There are certain things we already do without human supervision, like counting the number of cars on roads; those are mechanical things. It is the life-and-death situations, like a drone with a missile deciding whether to hit a target, where it is better to have humans in the loop.”
Addressing challenges; reaping rewards
Admittedly, along with a plethora of possibilities, AI also carries its own set of challenges. Take the spread of false information, for example. With the latest neural networks, AI can synthesise videos and news clips that are indistinguishable from the real thing3.
Even as policy makers and the tech industry take steps to regulate AI, it’s worth noting that in some cases, doing so may involve the use of more AI. “The only way we can counter synthesised news is to have another AI that can spot fake videos. I believe some groups are already starting to develop defences or mechanisms to filter out fake news or fake content,” Dr Tan said.
While time will tell how significantly AI will affect our careers and our lives as a whole, one thing is certain: there’s no stopping the changes to come. Rather than fear or resist change, we should learn to use AI wisely and reap the rewards it can bring.
“This is the most exciting development in human history,” said Dr Tan. “It brings new functionalities that previously, people wouldn’t even fathom were possible.”
1. Weng et al. (2017) Can machine-learning improve cardiovascular risk prediction using routine clinical data?
2. Ministry of Communications and Information (2017) 5 reasons to use the Gov.sg bot, https://www.gov.sg/news/
3. BBC News (2018) Tech Tent: Fake videos stir distrust, http://www.bbc.com/news/technology-43838310