The Long View: What ITC Vegas Said About Insurance Beyond the AI Hype
AI dominated the agenda at ITC Vegas. The sessions worth remembering were about the forces that will outlast it.
- Examples of the slow forces under the AI story: longevity, a K-shaped economy, extreme weather and eroding trust.
- How Mapfre and Nationwide think about planning further out.
- Why the last two years of losses mislead, and which risks they miss, from GPU residual values to cloud downtime.
- What gains value with time: data with history, decisions you can backtest and a continuous view of the outside world.
Ask a conference room about AI and you usually get this quarter's roadmap. At ITC Vegas, the sessions that stuck did something less common: they talked about timelines, not tools. They asked what the industry will look like when today's decisions come due. Read together, they described one habit in five moves.
1. The Real Story Is Slower Than the Hype
Step back from the headlines and the forces that matter are slower, older and already visible.
A fireside chat from Nationwide stepped away from the hype cycle and framed AI as "part of a bigger story". It named four long-horizon forces:
- Longevity: over 10,000 Americans turn 65 every day.
- A K-shaped economy: the bottom half holds around 4% of US wealth, the top 10% over 60%.
- Extreme weather: catastrophic losses keep rising.
- Eroding trust: one 2025 Gallup survey found only about 20% of respondents trusted businesses.
None of these arrived with a chatbot. They are the slow currents under every product decision in the industry and Trendtracker has been picking up these signals for years, long before they reached the conference stage.
If these slower forces are the real story, how far out should you plan?
.avif)
2. Plan Backwards From 2040
Start from where the customer will be, not from where the budget cycle ends.
Carlos Cendra Falcon, Scouting & Investment Lead from Mapfre, said their strategy is aimed at 2040 and 2050 more than 2030. The goal is not just to sell a policy but to be present wherever people are willing to engage.
The same Nationwide fireside chat turned the long view into a method: assess the trends at one, two, five and ten years out, turn them into specific customer problems. And it was candid about the limits of prediction:
"The goal is to not predict the future perfectly... build capabilities that respond to evolving needs and respond as the future becomes clear."
- Helene Wirth, Director Innovation Execution, Nationwide Insurance
But planning that far out only works if you stop treating the recent past as a forecast.
.avif)
3. Yesterday's Losses Are Not a Forecast
AI is moving faster than any risk the industry has priced before.
An insurer on the AI liability keynote put the warning bluntly:
"It would be a mistake to take the last two years of losses and project them forward, because coming capabilities are fundamentally different."
- Lynn Thompson, Global Head of Strategy and Partnerships, QBE Ventures
The pace is the problem. At the AI liability keynote, Swiss Re's Ali Shahkarami pointed out that cyber took more than a decade to mature, while AI tools spread across organizations in one to two years. Mapfre's Carlos Cendra Falcon added that almost nobody saw ChatGPT coming a few years ago, so we should expect more surprises:
"Lots of plans that were set up 9 months ago have completely changed due to rapid innovation on the AI landscape."
- Ben Battle, Director of Data Science, Accelerant
If the past is a weak guide, the map of what needs insuring is changing too.
4. The Risk Map Is Being Redrawn
New exposures are arriving before the old ones have finished changing.
Thompson urged a deliberate, forward-looking view at the same AI liability keynote: look at the risks clients actually care about, such as off-take agreements, GPU residual values, SLAs and cloud downtime, not just errors in model output.
Even the old lines are shifting. The same panel expected AI exposure to seep into cyber, professional liability, D&O, E&O and EPL, and rejected the easy way out:
"Blanket exclusions lack longevity."
- Lynn Thompson, QBE Ventures
So if the past is a weak guide and the risk map keeps moving, what is worth building now?
5. Build the Things That Compound
Invest in what gets better with time, starting with how you read the world around you.
Examples from the sessions share one trait: they get better the longer they run.
.avif)
Data with history. A Delos panelist at "From Data to Advantage" explained why money alone can't shortcut two decades of proprietary wildfire data. You need longevity, trends over time and real outcomes:
"In the end, you are trying to predict what's going to happen in the future. And that doesn't work if you're overfitting to a very limited sample size of situations."
Decisions you can learn from later. At "From Insight to Impact," an AI agent reviews contractor documentation. The saved results can be backtested over three to seven years as claims develop. That shows which contract requirements really mattered and which were mandated for nothing.
A view of the world outside. These examples share a trait: neither could be reconstructed after the fact. The same holds for how an insurer reads the forces around it: a record of what was moving, how fast and on what evidence only becomes useful once you have been keeping it for a while. That kind of watching can't be a one-off project. It has to run continuously, with an evidence trail you can return to. That is the difference between a snapshot and a view that builds over time.
.avif)
From Reacting to Anticipating
The advantage goes to those who take a long-term view and recognize slow signals early.
Side by side, the pattern is clear. AI got the airtime, but the most useful conversations took the long view: longer horizons and assets that compound.
That makes the useful questions about foresight, not technology. Which of these forces will shape your customers in 2040? What are you still pricing off the last two years? What are you building that gains value with time?
The answers sit in signals outside your walls. Start with our Megashifts report, which maps nine forces into scenarios your team can act on, or talk to an expert about tracking them continuously, so you see the slow signals before they become next year's headlines.
