Risk & Robots

ChatGPT often hallucinates, making it unreliable for reading insurance policies, which demands high accuracy. The Retrieval-Augmented Generation (RAG) offers a solution by providing AI with access to a knowledge base to reference accurate information. However, implementing RAG is resource-intensive, as it involves gathering and labeling vast amounts of precise data. This is a video short of an interview Pankaj Parashar & 🇺🇸 Nick Lamparelli did with Dan Schuleman, CEO of Qumis. Full episode to drop next week #RiskandRobots #AI #Insurtech

You can now use ChatGPT within Excel and Google Sheets No more memorizing complex formulas or functions Write killer prompts, generate formulas, categorize data and more Here’s how you can do it: • Open Excel and click on Insert → Get Add-Ins • Search Numerous and Add Numerous Add-In • Open a spreadsheet and write your prompt using =AI(Write your prompt here, cell reference) formula Check it out now: https://numerous.ai/ __________ If you like this, and want more AI tools, tutorials, and news, join Superhuman, my AI newsletter with 600K subs now: https://lnkd.in/dXQ9-B9A

Ok, anyone read the #Noahpinion post this morning about AI (by Noah Smith, whose writing I adore)? It was a much clearer layout of something I have long believed - AI may well make our humanity more valuable, not less (working at the top of our licenses, in insurance-speak). It’s comparative advantage - no matter how good AI gets, it’s always going to make sense for humans to do the things humans do best relative to the other things we can do. There’s no reasonable world in which AI takes it all over, because there’s a always a limit on energy available to pump into AI computing, no matter much more efficient it gets/how much more energy we can access. What’s our comparative advantage? Connecting with other people, empathy, understanding, connection. Seeing connections that haven’t existing before. Breaking the fourth wall between data and the human activities that create the data. All things that computers can’t do. Heck even social media doesn’t do those things, and that’s human created content sitting on tech platform. What kinds of tasks are likely to matter more, not less, in insurance in an AI world? I’m no futurist, but here’s what I’m thinking. What else? 💡 Risk assessments/reassurance/counseling for individuals and businesses 💡 Return to work support after a workplace injury 💡 Training AI models to understand the humans that create the data 💡 Management, generally 💡 Bus dev, M&A, and similar They’ll all be hyper charged by AI tools, but the most important parts of these jobs will be even more valuable. Also, here’s what DALL-E comes back with for “Draw a picture illustrating comparative advantage for humans vs AI.” I like where it’s going. #aiininsurance #insurance #insurtech

"First rule of engagement: AI's not in charge, humans are. We can delegate responsibility to AI, but never accountability."

Let's talk about handrails, underwriting..........and Generative AI. 😎 I drive past this house on a weekly basis. It's a nice place. Sits on 5 acres in rural northern Michigan. Big garden. Beautiful country setting. Stunning views. 🏡 🌤 The house was built in 2018. It's a 1.5 story, frame construction home, with a big, elevated deck on the front. Like EIGHT FEET elevated..........with no handrail...........at all. 😨 Zero. Zilch. Nada. And it's been this way for 5 years. 🗓 It's what I refer to as an "Umbrella Deck". ☂ Because when someone falls off the deck and breaks their neck, I hope and pray the homeowner has an Umbrella Liability policy...........because $1M of liability coverage on their HO-3 ain't gonna protect their assets in court. 💸 But wait.......there IS a handrail, you say? OK, I admit it. A handrail is visible in the photo. Looks legit and safe. ✅ Buuuuuuuuuttttttt........check the FIRST comment below. 👇   The handrail IS NOT REAL. It doesn't exist. 🤯🤯🤯 A few clicks and a simple text prompt of "add handrail", and taaaa-daaaaa! Liability eliminated in under 20 seconds! Woohoo!!! 😑😑 This is Generative AI. And it's a BIG problem for insurance underwriting and claims. 🛑 I'm excited about what we're doing to ensure our self-inspection photos are legit and have not been digitally altered with GenAI or other photo manipulation tools. 🤳 Because its become WAAAAYYY too easy for bad actors to game the system. And underwriters must have confidence that the photos they're looking at are an accurate depiction of the real-world risk they're analyzing. 📷 #underwriting #claims #GenAI #fraud #insuretech