Artificial intelligence (AI) remained top of a compelling agenda on Day 5, the final day of Hong Kong Fintech Week organised by Invest Hong Kong (InvestHK), one with keynotes from Zhong An, Swiss Re, Pintec and Ping An Technology amongst others on subjects including insurance, loan management, robo-advisory and the future of financial services.

Arranged by Finnovasia, the Friday sessions covered AI and deep learning, which inevitably led the conversation into the future of financial services, including the impact of AI on jobs. There was a clear consensus that actuaries, fund managers, financial advisors, insurance agents and others will need to develop new skillsets down the line. The sector has already seen transformation, but there’s a lot more to come.

The opening speakers in the morning session on “How AI and Machine Learning are Forever Changing the Insurance Industry” drew another large audience. First up was Ericson Chan, CEO of Ping An Technology, in which he reiterated the need for AI to be made relevant to customers, while emphasising the importance of data in achieving that.

“Thanks to Alpha Go for putting AI on the map, but who plays Go? I did maybe 30 years ago,”

he said jokingly to make the point:

“AI needs to be relevant.”

He referred to more tangible benefits that customers can relate to: practical applications that are faster, cheaper and easier to use.

Ping An is China’s largest financial institution that isn’t a state-owned enterprise. It has more than 400 million online users and invests 1% of revenue in research and development – about US$1 billion a year. It has around 20,000 R&D staff and has the highest number of patents in China.

In dealing with that volume of business, use of facial recognition technology has proven to be an important measure for risk management. It includes expression analysis technology that can detect confusion, nervousness, embarrassment and so on, which helps direct the line of questioning – perhaps the customer needs more explanation; perhaps mischief is afoot. All of this can be worked through a mobile app video conferencing tool.

Such technology facilitates the know-you-customer (KYC) process.

“It goes beyond spotting terrorists and money launderers,”

Mr. Chan said.

“It’s about really understanding the customer in order to make recommendations that go beyond pure wealth management.”

He said this would include suggestions on how the customer could become healthier, safer and so forth.

“We have 20,000 data labels per customer.”

The key point made by Wayne Xu, COO of Zhong An Insurance, China’s first online insurance company with 500 million customers, was that

“AI is going to change the whole of society, not insurance or one particular sector.”

He pointed to healthcare, wealth management, autonomous vehicles and public security as examples of some of the areas in which AI will be transformational.

With its customer base and 8 billion policies underwritten since it started in 2013, Zhong An has generated a vast amount of data, which AI uses as part of the company’s basic infrastructure. Mr. Xu said there are three main categories of AI in insurance: prediction, which helps with products and pricing; communication, such as chatbots, which improve the quality of service; and identification, to both identify customers and identify fraud.

He gave examples of AI usage. In customer services, chatbots go first when dealing with customer enquiries, followed by humans if the chatbot doesn’t have the answer. However, all the human answers are saved to a database, so they can be used by the chatbots in future. He said that 97% of questions can be directly answered by the machine, without any human intervention.

In the future, Mr. Xu said, an insurance company could run by itself, with just five or ten people running it, and “to talk to the regulator.”

Kin Tse, Data Scientist, Digital & Smart Analytics at Swiss Re, talked about some of the practical applications of machine learning in the risk underwriting process.  He provided an interesting example in the context of health profiling.

“After age and gender, smoking is the number one risk factor in profiling,”

he said.

“But between 20% and 50% of smokers do not disclose their smoking in their (insurance) applications.”

In short, they lie. Such deception undermines the underwriting and creates risk. To counter it, Swiss Re has created a model to identify people most likely to be concealing their smoking, so that they can be tested further.

“We have a 90% accuracy rate,”

said Mr. Tse.

“We’re been able to bring down fraud rates from 50% to 5%.”

All the companies that presented during the week offered distinctive business models and areas of expertise. But a standout from Hong Kong was SenseTime, described as the world’s biggest AI unicorn following funding round earlier this year that raised a huge US$410 million. Its Managing Director Shang Hailong described the company as Asia’s only deep learning platform.

“AI providers provide applications, but SenseTime provides the only deep learning platform. The only 100% core technology platform in Asia,”

he said, illustrating some of its capabilities with practical examples, including vehicle damage assessment.

Hong Kong featured strongly in a panel session arranged by Shanghai-based media company Yicai, which kicked off with a speech by Charles d’Haussy, Head of Fintech at InvestHK. He reminded the audience that Fintech in Hong Kong is primarily business-to-business. He added that growth of the sector has been organic and driven by the private sector.

“Hong Kong has space for disruption, but it’s more about financial services innovation,”

said Mr. d’Haussy, with regard the city’s established position as a regional financial hub with 75 of the world’s largest 100 banks operating there with full licences.

“People value maturity of regulations and stability, which is what Hong Kong offers,”

he said.

“There are trade-offs, but there are great advantages which are very important for Fintech companies.”

Mr. d’Haussy pointed out that Hong Kong now has five Fintech accelerators, none government sponsored.

“All the regulators have infrastructure that welcomes innovation in Fintech,”

he said, reminding the audience about the three new regulator sandboxes announced last month – for the Hong Kong Monetary Authority, Securities and Futures Commission and Insurance Authority. Details of these can be found at sandboxes.hk.

The huge opportunities for AI in the financial services space were laid out by Jason Wang, Head of Strategy (Financial Services) at Baidu. He said there are 800 million active investors in China, of which only 300 million have a financial situation that’s known, meaning the remaining half billion cannot readily get access to loans. He added that 50% of SMEs cannot take out loans.

The imbalance and need are clear, said Mr. Wang.  With non-traditional profiling using data still more than 80% accurate, people can be given a risk profile, he said. This would help address the need.

Mr. Wang added:

“10% of banking (in China) is online, 6% loans are and 1%-3% of insurance is done this way. We won’t reach 100%, but there’s room for growth…it’s a great opportunity to lower cost and increase efficiency.”