DWIC18 - New data and processing power make way for “renaissance” in AI

Artificial Intelligence (AI) has been touted to have an “exponential impact” on the industry, but panellists warn that it has to be properly supported by the overall business strategy in order to be successful.

In the first panel session at the Dubai World Insurance Congress 2018, hosted by Global Reinsurance, technology experts from Deloitte, IBM and Union Insurance unanimously agreed that to implement AI into a business, it needs to be driven from the top and carefully positioned in the wider strategy.

On common mistakes businesses make when looking to position AI into processes, Nigel Walsh, partner at Deloitte said: “We collectively believe that AI solves everything without homing in on something small or manageable enough to prove its success quickly in the first instance.”

To successfully embrace AI, leaders should be aware of what business needs AI can fill, said Walsh. “Businesses need to start with the ‘why’ and be as specific about this as possible as this message will eventually be communicated to other employees who need to get on board too,” he said.

“Set clear, realistic objectives,” he said. “The establishment of an AI organisation is organisational change, more than just a technological change,” Walsh alluded to in a presentation opening the panel.

IBM’s Bashar Kilani talked of a “new era in technology” which is going to have a huge impact on the way businesses and people access and process data.

“90% of the data we have access to now has only been created in the last two years,” said Kilani. “The processing power we have today has never existed before in this way, so when we put this processing of power and availability of data together we have a renaissance in AI.”

“Those that can act quicker, will no doubt be winners in this space. Can companies afford to ignore this? Absolutely, categorically not.”

Nigel Walsh, Deloitte

Anshul Srivastav, chief information officer and digital officer at Union Insurance, said: “AI has been around for years, but we didn’t value data in the way we do today.

“We didn’t value the need to become more efficient. We didn’t explore properly how we can improve customer interaction through technology, but now this is much higher up on the agenda,” said Srivastav.

Kilani added: “The insurance industry is very well placed to take advantage of the developments in both data and processing.

“AI will touch every aspect of the chain from efficiency through to customer interaction. We will move into a world where a lot of it is done in the background by AI, in addition to all the other information, intelligence and data the process will be able to collate along the way.”

But does this “new era of technology” instil fear within the industry, namely around the job security? Deloitte’s Walsh doesn’t believe this fear exists anymore.

“I think we’ve gone past this point quite quickly and now we’re looking at how it’s going to benefit us and our ability to differentiate through being able to get faster insights on customers, the risks that they carry,” he said.

“Those that can act quicker, will no doubt be winners in this space. Can companies afford to ignore this? Absolutely, categorically not,” said Walsh.

Kilani added: “AI will be underneath every business process, every interaction. We expect 90% of human mistakes to be removed by AI.”

But Srivastav raised skillsets amongst employees as a potential barrier to successful implementation and one the industry needs to get better at addressing.

“Despite how far we’ve come, data and analytics is still taking a back seat. A huge amount of skilling and reskilling is required to ensure it’s being valued to its greatest potential,” said Srivastav.

Adding to the relevance of “humans” in this process, Kilani said the foundations of AI are built on “man and machine”.

“Augmenting the intelligence of data analysts and taking it to the next level starts with man and machine. The AI systems and processes which will determine the future of businesses - both inside and outside of the conversation around insurance - are categorised by two things: what data has been used to train the system and who trained it?”