China’s biggest tech companies are all diving into the business of insuring motor vehicle accident damage. Using A.I. – artificial intelligence – to provide protection against car accidents may seem marginal to financial services, but it is the vanguard for China’s ‘techfin’ companies to reshape insurance, investments and banking in their image.
Of course these tech companies have been involved in insurance and other financial services such as payments, but the sophistication gained by combining different forms of A.I. is taking their offering to a new, expansive level.
Drivers take pictures and video of their damaged car at the site of an incident. They take selfies to authenticate their identity as the car owner or driver. The tech company ensures the make of the car and the license plate number match their record. Their computers assess the damage, and provide itemized lists of problems, the costs to fix them, the amount of coverage available, and a map with nearby garages. If the customer agrees to the cover, the money is in their account within a few minutes.
Data, scale, speed – and ambition
These apps not only let customers claim their auto insurance quickly. They also help insurance companies cut back on the need to send employees to accident sites. Some car crashes will require human intervention, and some types of damage are hard to analyze based on photos alone. But the bulk of routine assessment can be now automated, and the customer can sort out the entire process vis-à-vis their insurer immediately.
Chinese tech companies say this service is just a flavor of what A.I. will deliver, provided they amass enough data to deliver a particular service. Although foreigners may fulminate about Chinese privacy law, the fact is that Chinese consumers are perfectly happy to share their data – and at a scale of hundreds of millions of customers, says Ericson Chan, CEO of Ping An Technology.
This willingness to share, combined with incredibly fast computer processing optimized for neural networks and machine learning, and with a conglomerate mentality among tech providers, is what sets China’s tech leaders apart.
Ping An Group, for example, has 32 subsidiaries, of which half are in financial services and the rest are in things like autonomous cars, healthcare and housing. “You can’t just do finance if you want to serve your customer holistically,” he said, speaking at Hong Kong FinTech Week. “What’s money for?”
It’s the same story for Alibaba’s Ant Financial and Tencent’s various financial subsidiaries, including WeBank and WeWealth.
New markets, lower risk
Douglas Feagin, president of Ant’s international business, says A.I. is the key to enabling a range of functions work better, from opening an account to selecting a service to managing risk. “We’re investing in machine learning, natural language processing, prediction, image recognition and speech recognition,” he said. “All of these things need to come together to create a more innovative form of financial services.”
Shang Hailong, Hong Kong-based managing director at SenseTime, says access to hundreds of millions of people’s data makes car accident insurance the easiest place for tech companies to apply A.I. to financial services. The company, which raised $410 million in a B round from venture capital in July, is working with mainland banks and tech companies to help them further develop A.I.-led initiatives in health and life insurance, as well as wealth management.
“A.I. works now because of deep learning,” he said. In turn it allows an entire business to be automated, enabling insurers to reduce their cost of human capital – and provide a better service at the same time. “Damage assessment is no longer subjective,” he said.
This technology also allows insurance companies to expand into new markets. Daniel Hong, deputy general manager of financial technology at Tencent, says automating everything from accepting new customers to assessing car damage allows insurance brokers and carriers to extend their offering to smaller cities, beyond the reach of brick-and-mortar offices.
Such a trove of reliable data doesn’t exist in other businesses yet. For example, Hong says robo advice today is too simple and prone to customer manipulation if they aren’t truthful in how they respond to questions about risk appetite. He says using A.I. to generate actively managed investment products is probably better suited to hedge funds than to retail customers – for now.
But Tencent is gaining enough user information via its WeChat messaging service (ubiquitous in China) that it is learning how to match wealth-management attitudes with actual consumer behavior.
And A.I. and deep learning are better at managing risk and detecting fraud. Alipay’s fraud loss rate is less than one incident in a million transactions, far below what other global payment players experience, says Feagin. “We do this by scanning thousands of variables in a given transaction,” he said, noting that the company is so confident it will protect consumers for losses up to Rmb1 million for a premium below Rmb2. “Everyone will opt for that,” he said – and it works for a company with over 500 million customers.
Tech is not a tool
China’s techfins are not erasing insurers or other financial institutions, but turning them into rather simple suppliers. Ant, for example, uses China Taiping Insurance to provide the actual motor-damage insurance product. But it is the tech companies who own the customers using their apps to access and use insurance – and it is they who get access to consumer data.
This is what makes this trend different from mere digitization of business on a nice app.
Wayne Xu, COO at Zhong An Online P&C Insurance, says A.I. represents a cluster of techniques that when combined, put computers in the driving seat.
Traditional insurance relies on actuaries to develop smart, statistics-based guesses around risk. They use computers to assist human judgment. “Tech is just a tool” for these industries, Xu said. “But with A.I., it’s the computer that makes the call. Humans just confirm it.” Tech is no longer just a tool; it’s the business.
Scale makes A.I. viable. Hundreds of millions of users provide meaningful data points that machines can learn from – and as information accrues, the machines will just learn it better, assuming they are being taught correctly. But even for a Zhong An, with over 500 million customers and 8 billion policies sold since 2013, it needs A.I. to survive. Only machines can process such volumes.
Moreover, companies can only acquire such vast numbers of customers by offering rock-bottom policies. Some Zhong An and Ant Financial premiums cost only Rmb1. Meanwhile, Alibaba has invested Rmb100 billion into A.I.-related research and development.
Selfies and karaoke
Ping An’s Chan says most A.I.-backed offerings in finance have involved a single function or type of tech, such as chatbots, peer-to-peer networks, and simple robo advisors. The competitive advantage is how to blend them – and to make them easy to consume.
While for techies, the killer ingredient may be facial recognition – it’s what has made car-damage assessment viable – it’s also about knowing how to sell a new service.
Taking lessons from car insurance, Ping An is starting to deploy this technology to other subsidiaries selling asset management and other insurance products.
Instead of sitting in a bank’s office for an hour going through a cumbersome application process, customers can do everything remotely with their smartphone. Instructions crawl along the screen – like the lyrics to a song during a karoke session – to guide both the sales agent and the customer through the process.
It takes ten minutes, but more importantly, it doesn’t feel like such a headache. “Whether you’re claiming car insurance or buying a fund, it’s just taking a selfie and karaoke,” Chan said.
This article originally appeared on October 31, 2017.