Four For Friday | January 24, 2025
LF157 | Are we in the end game of insurance?, Meta's new glasses, OpenAI's new longevity model, Liverpool's new AI healthtech plans, + Bonus: DeepSeek
Welcome to Looking Forward’s Four For Friday. Four things that have piqued my interest this week, together with a bonus: AI Tip of The Week. Enjoy!
Are California’s wildfires the end of insurance?
A cheery note on which to start this week’s newsletter… Financial losses from the recent LA fires are estimated at more than $250bn, but the bigger issue is whether this is going to spark a wider crisis in insurance itself, and even a possible financial system contagion.
As insurers start to withdraw, the value of the homes that can’t get insurance plummets, with knock-on effects for the economy. Plus the insurers have a smaller base over which to spread their premiums.
Anyone who remembers the Great Recession of 2007-2009 knows that seemingly localized problems can snowball. In that event, the value of opaque bundles of real estate derivatives collapsed from artificial and unsustainable highs, leaving millions of mortgages around the US “underwater.” These properties were no longer valued above owners’ mortgage liabilities, so their best choice was simply to walk away from the obligation to make their monthly payments. Lenders were forced to foreclose, often at an enormous loss, and the collapse of real estate markets across the US created a global recession that affected financial stability around the world.
The So What? Thinking that the future is going to be like the past is creating massive problems for insurance, and potentially the broader economy. We need more robust scenario planning, new cross-industry collaborations and out-of-the-box thinking to address existential risks.
Meta developing Oakley athlete glasses and camera-buds
I’ve been pleasantly suprised with the Meta Raybands I bought a few months ago, while stumbling across Meta’s corporate shop during a ramble around Foster City. The picture, video and audio quality is impressive. It’s great for playing with the kids (specifically throwing them in the air, and then (mostly) catching them); sometimes having your phone in your hand is downright dangerous.
Meta’s AI features, on the other hand, have been pretty useless for me - it either doesn’t know the answer to simple questions (Where am I?) or says it can’t tell you (perhaps due to people freaking out about the facial recognition use cases).
So am following this story about Meta’s new product evolution: an athlete’s glasses line Oakley, a smart watch / strap to control glasses, glasses with an in-built computer (so no smartphone needed) and a new line of Airpod-style earbuds with cameras in them.
The So What: Hardware is hard, but as Apple showed us, tight vertical integration with hardware and software (now AI), can make for a great user experience, and margins.
OpenAI develops AI model for longevity
Sam Altman has already put $180m into Retro Biosciences, and now OpenAI has launched a new AI model focused on longevity science.
Retro has been focused on ‘Yamanaka factors’ to make it easier to convert skin cells to stem cells (which can then help reprogram the rest of the body). OpenAI’s new ‘GPT-4b micro’ model was used by Retro to boost the effectiveness of these factors by 50x, thereby potentially accelerating new discoveries.
As an aside, I found this story via CB Insights, and here’s their ‘longevity tech market map’, which frames the tech-longevity space (and has Retro part of the ‘cellular and epigenetic reprogramming’ category.
The So What? Watch out for new vertical AI stacks where models and data are optimized for specific solution areas rather than being general chatbots.
Liverpool’s new Healthtech Innovation Zone
I’ve had the pleasure of collaborating with the folks at Civic Health Innovation Labs (the rather delightful and appropriate acronym, CHIL) in Liverpool over the past year, and so it’s good to see they look set to now land an initial £10m funding to start to build out some of their very bold ideas about using citizen-focused data and AI to improve social care. They’re building a local data cooperative that aims to ensure the citizens creating and sharing their data will be able to own and benefit from it.
Although a rounding error compared to the numbers coming out of the US (half a trillion here, half a trillion there, and it soon adds up), or even the new UK AI plan with £14bn in corporate commitments), I would bet on these guys to be the first to get an ROI on delivering AI solutions, that both create jobs, support local businesses and improve the quality of health and social care in the region.
The So What: Don’t write off universities; this kind of citizen-centric activity is only possible with the trust that their reputations and deep local integrations provide.
Bonus AI Tool: DeepSeek
I’ve been playing with the new Chinese-created AI model, DeepSeek, and have to say on many of the tasks I’ve been giving it, it just feels better, than OpenAI or Claude. Apparently it was trained on 15 trillion words, but only cost $5m to build. Makes you wonder what OpenAI needs $500bn for…
That’s all for this week. As always, feedback welcome. Feel free to share insights or links of interest.
- Stephen