Generative AI in Practice is InsTech’s monthly newsletter dedicated to the use of generative AI in insurance.
The InsTech Perspective
Are you ready to be surprised?
“Because a large language model is provided on a computer, people treat it like a computer application and an algorithm… Organisations need to start thinking about this differently. If you want the power of it, you want to be surprised by it. You want it to tackle situations that you didn’t test and you want it to tackle those in the right way. So you have to have the controls in place to understand when it has not done what you wanted it to do.” – Colville Wood, Chief Technology Officer for Insurance (UK&I and EMEA), Cognizant on InsTech podcast episode 283.
Most technology is ‘deterministic’; it operates under a predefined logic and produces consistent and predictable outcomes. Add a contact to your CRM, and it appears as expected. Query your policy management system about a company, and the details pop up without fail. Even the machine learning algorithm that calculates your car insurance premium is deterministic; feed it the same data and it will produce the same results.
Generative AI is different. In some ways, it behaves more like a human employee than a deterministic algorithm. It will produce creative outputs for weird or complex problems where other technology would be unable to respond. Generative AI’s inherent unpredictability is what makes it so capable for various tasks. It also means any generative AI application needs robust oversight (similar to a new employee) to make sure it is doing the job correctly.
What the insurance industry is doing with GenAI
Aon plans to roll out a set of generative AI tools to its employees over the coming months. One use case is comparing two policy wordings received from different insurers; the AI tool will highlight the differences.
Retail broker Newfront is using generative AI to communicate with insurers and wholesale brokers, while wholesaler CRC is looking to automate how it matches risks with the right underwriters. Learn about these and more use cases by listening to our recent podcast: the brokers using generative AI today, with CRC Brokerage’s President and Newfront’s CTO.
HDI Global has launched a generative AI tool to help its staff extract information from unstructured data including legal contracts and claim documents.
Reinsurance Group of America (RGA) has invested $12 million USD into start-up DigitalOwl, which specialises in analysing medical records using GenAI. RGA is integrating DigitalOwl’s technology into its systems.
QBE Ventures has invested in Snorkel AI, a company helping large organisations customise large language models.
Norwegian MGA TT Forsikring is using Simplifai’s solutions to handle customer inquiries and manage policy cancellations with generative AI.
Maine Employers’ Mutual Insurance Company (MEMIC) is using a GenAI tool from CLARA Analytics to analyse medical records and save time for its claims staff.
Insurer Foyer Global Health has adopted a GenAI tool from Galytix to triage incoming claims.
MGA Vouch has launched an insurance product covering AI companies against lawsuits associated with AI errors, discrimination and intellectual property disputes.
News from the InsTech network
QBE Ventures, Miller Insurance and Inver Re (Ardonagh Group) will be discussing generative AI in an InsTech webinar on 28 March. Now that many pilot projects have been tried, what are the proven use cases? Register here.
Claims technology company Synergy (part of Claims Consortium Group) sets out where it is focusing its AI efforts.
QBE Ventures’ CEO discusses QBE’s perspective on GenAI and other areas where technology investment can drive strategic value to insurers.
30% of insurance and financial services firms invested $50 million USD or more in AI in their last fiscal year. 36% have implemented company-wide AI initiatives, while AI is used in some specific business functions by 55%. Read all the outcomes of EXL’s survey here.
Indico Data is hosting a webinar on 6 March to discuss how insurers are using LLMs and automation to drive better decisions, with speakers from QBE Ventures and Everest Insurance.
More than 75% of insurers plan to use AI to automate tasks in the next three years, but there are challenges. Eigen Technologies writes about the roadblocks to automation in insurance and how to overcome them.
What types of tasks is generative AI well suited for in insurance? Cognizant’s Colville Wood and David Fearne join the InsTech podcast to discuss the power and surprises of GenAI in practice.
Other news and insights we’ve seen
Over 100 insurance professionals have collaborated on a ‘voluntary code of conduct for the use of artificial intelligence in claims’. Its early signatories include Covéa and Ecclesiastical.
The British Insurance Brokers’ Association (BIBA) is calling for the UK government and regulators to introduce a framework for the regulation of AI in insurance. BIBA says it will also produce guidance for its members on how to benefit from AI.
The founder of a financial planning start-up writes that generative AI will empower consumers. “Take a robo-caller, add generative AI and direct it to negotiate [a claim] with your insurance company… While those who can afford it may want a [lawyer], for the majority of consumers who can’t afford to fight, or don’t know how, an AI-powered advocate is a godsend.”
How InsTech has used AI this month
ChatGPT is rather like an intelligent intern. Whilst it can perform various tasks and answer questions generically, it has no specific knowledge of your organisation apart from what you tell it. As a result, companies have started using a technique known as ‘retrieval augmented generation’ (RAG) to create chatbots that answer questions based on a body of knowledge. Connect it to company policies and it becomes an HR expert. Connect it to files, documents and notes and it acts as a personal assistant, finding the information you need quickly wherever you saved it.
At InsTech, we’ve been working with RAG specialist myReach to create a chatbot that answers questions using all of InsTech’s recorded knowledge: podcasts, articles, reports, event recordings, newsletters and more. When it gives a response, it also shows which resources it used to answer the question. See below for an unedited example of the chatbot in action – soon we plan to make it accessible to members on our website so you can try it out too. Find out more at myreach.io.
Find out what you missed
Issue 6: AI shortages, algorithmic underwriting and ChatGPT podcast takeover
Issue 5: In-house chatbots, AI in healthcare and designing reports
Issue 4: Catastrophic risks, insuring AI and pet insurance insights
Generative AI in Practice is InsTech’s monthly newsletter dedicated to the use of generative AI in insurance – you can sign up for free here.