No. 5, May 2023
Insurance data and technology commentary and news from Quotech Founder Guillaume Bonnissent.
Chat backchat
ChatGPT v.4 was released at the end of March, to much fanfare, adding yet more fuel to the seemingly endless stream of hype about generative AI. It’s the branch of artificial intelligence which creates text, images, or other media in response to inputs. It powers, among other things, the “chatbots” one must satisfy when seeking tech support before being put through to a thinking human.
One industry commentator, the expansive MGA CFC, said: “There is no wonder [Chat AI] is generating a huge amount of interest with its potential to change how businesses operate and how people live their lives.” But will AI be truly transformative?
Such statements should always raise an enormous red flag. Think back to claims that blockchain was set to revolutionise the reinsurance sector. When all the hype finally died, blockchain turned out to be just another accounting software system, albeit a new generation of a familiar tool (the revolutionary account-book). As Insurance Business reports, “It’s unlikely, at its current stage, for generative AI to completely replace underwriters, claims handlers, or customer service representative.” Well, that’s a relief.
The simple truth about the value of generative AI technology is the same as it is for all other tech innovations. It is another useful tool among many which will help to automate processes, solve problems, and improve efficiencies in the insurance sector.
We must start with real problems, not with trendy solutions. Example: a Quotech client’s D&O underwriters spent hundreds of hours per year reading potential clients’ 10-K forms. We created an AI tool which does that within the underwriting desktop system they had commissioned. (That was long before the excitement about ChatGTP.)
It may not be the approaching revolution, but I do believe insurance-company executives should spend time thinking about generative AI technology. That’s because it will almost certainly lead to claims in a number of classes, including PI, E&O, D&O, and cyber. After all, we all make mistakes, including our silicon friends – and where there’s blame, there’s a claim.
MGAs AND TECHNOLOGY
It’s good to see our old friend William Pitt, who these days is a director at Conning, declare the need for better deployment of technology in the MGA community. In an article in The Insurer, the Marsh, Beazley, and Whitespace alumnus declares that “most MGAs still have to raise their game in exploiting technology.”
He goes on to highlight several of the many MGAs which do so. His list includes Kettle, reThought, Pie, At-Bay, and Coalition – many of the new-fangled “insurTech” MGAs. But MGAs with a much more traditional approach can also make it in the space – in fact, their chances may be even better.
Most insurTechs have some whiz-bang platform that helps them underwrite better or more efficiently. Few start with a market share, broker following, sector-leading product, or any of the other assets that count in the traditional MGA space. That means an existing MGA that has at least one of these could gain a long-distance jump on its upstart insurTech competitors with a systems retro-fit. At Quotech, we’ve made it happen.
Elsewhere in the news…
Some data stories that caught my eye:
Use of the OED (the Open Exposure Data Standard, not the hallowed dictionary) has taken a step forward with news that Aon, RenRe, SCOR, Hannover Re, and Swiss Re, are testing its efficacy. The OED has been around for years. It was set originally by the tech services firm Simplitium, which was later acquired by NASDAQ, and has since been placed in to the hands of the OASIS Loss Modelling Framework (who recently, usefully, added a cyber standard). The news is a positive sign of the market’s emerging willingness to move away from proprietary data formats.
Jay Borkakoti, currently Head of Data Products at the B2Small-B broker Simply Business, was the latest guest on the always-insightful Wenalyze webinar Insurance Is all about Data. He asked viewers: “Can you ingest a real-time data set in less than ½ a second?” Most cannot, but all could. QBE Group CEO Andrew Horton said something similar on Mark Geoghagen’s excellent The Voice of Insurance podcast. “A lot of companies don’t use the current data they’ve got,” he says. “I want to focus on that.” Andrew, give me a call, Quotech can help!