🔮 The Sunday edition #506: World on fire; brain simulation; Meta’s choice; home supercomputer, smart ants & rodent VR ++

An insider’s guide to AI and exponential technologies

👋🏼 Hi, it’s Azeem. As Los Angeles grapples with devastating wildfires and global temperatures breach the critical 1.5°C threshold, we face stark reminders of our changing world. The insurance industry’s retreat from climate-vulnerable regions signals a broader crisis of adaptation. Yet, alongside these global challenges, we’re seeing remarkable advances: scientists have achieved a breakthrough in brain simulation, matching the scale of human neural networks with 86 billion digital neurons. Today’s weekend edition explores how we navigate our planetary boundaries and technological frontiers.

The uninsurable frontier 

Wildfires tearing through Los Angeles are on track to be among the costliest in US history. One natural disaster after another begs the question: who will bear the cost?

Lex Sokolin

estimates that although $50-60 billion in damage looms, only $10 billion of private coverage plus $25 billion from California’s FAIR Plan stands ready – leaving tens of billions unaccounted for.

Adding to this misalignment is Proposition 13, which locks in artificially low property taxes until a sale triggers reassessment. Recent research finds that once a wildfire forces turnover – through destroyed homes and rebuilding – local governments can see a 10.5% overall revenue boost within five years, including a 21.2% jump in property tax receipts. On paper, that surprising surge offsets some short-term losses. However, it underlines the broader irony: local authorities may inadvertently reap fiscal gains from periodic destruction, while homeowners and insurers bear the brunt of rebuilding costs.

Meanwhile, Proposition 103 still largely prevents insurers from using forward-looking climate models, forcing them to rely on outdated or incomplete data. Though California regulators have begun easing that rule, the real impact is yet to be felt. In the meantime, insurers will continue to withdraw from fire-prone zones rather than write unprofitable policies, leaving the FAIR Plan (effectively the taxpayer) as the insurer of last resort.

Taken together, these top-down constraints distort prices and dampen incentives for more robust risk mitigation. If insurers were allowed to price fire risk fully, prospective homeowners in at-risk areas would face higher premiums and either build to higher fire-safe standards or move to safer zones. 

See also: Climate-tech investor

Yishan Wong

provides more insight into the home insurance market in southern California, including the $105,000 deductible his policy mandates.

Matrix brain

A new platform can now simulate 86 billion digital neurons and 4.76 trillion synapses – roughly on the scale of an entire human brain – by tapping more than 14,000 GPUs to model up to 47.8 trillion synapses. Researchers call this the “Digital Brain” – a mesh of spiking neuronal networks calibrated using MRI, diffusion-weighted imaging and PET scanner data.

From Simulation and assimilation of the digital human brain

The engine behind the model enables the simulator to approximate blood-oxygen-level-dependent signals observed in real fMRI. That correlation coefficient of 0.93 at a voxel level is particularly impressive. The simulated brain was also moderately good, with a correlation of 0.575, at predicting whether the subject would find an image pleasant or unpleasant.