โGOVERNANCEโ
A worthy successor earns power through consent.
A safe successor waits for humans to grant authority.
AI that canโt refuse power betrays alignment.
โCAPABILITYโ
ฮทโ = c(-ฮL) / ฮE
More learning per joule of compute
Capability without accountability is instability
Cool research but the headline jumps ahead a bit. Optical compute has huge potential for parallelism and low energy cost, but weโre still a few steps away from replacing GPUs. Most likely weโll see hybrid photonic accelerators first, not full optical AI chips. Still a real signal to watch.
Worth paying attention to.
Private comms keep systems stable by distributing coordination.
As AI ramps, centralized messaging becomes a single point of failure.
Reinforce it early or deal with fragility later.
TL;DR secure messaging had better survive the โAI ageโ or we are well and truly fucked.
Thereโs a reason comms networks are the first target of a military siege: because theyโre that important to everything else.
I finally reached human-level performance (85%) on ARC-AGI v1 for under $10k and within 12 hours. I use the same multi-agent collaboration with evolutionary test-time compute, now powered by GPT-5 pro with lower parallelism.
(1) Our team at @GoogleDeepMind has been collaborating with Terence Tao and Javier Gรณmez-Serrano to use our AI agents (AlphaEvolve, AlphaProof, & Gemini Deep Think) for advancing Maths research. They find that AlphaEvolve can help discover new results across a range of problems.
Over the last few months, my team and I built a dataset of 500+ planned data centers in the US with a combined capacity of ~125 GW.
Here's how we found them:
๐งต
Handing frontier compute to a rival might seem insane. The real question is timing.
Downstream actors take the 1st hit: bureaucracies, industries, allies. Some shocks donโt reverse; they reshape the field.
If adaptation outruns lost leverage, stability returns leaner & faster.
I would simply not give away the essential bottleneck input for the most important dual-use technology of our era to the US's primary geopolitical rival.
@janet_e_egan Energy is the substrate.
Compute without generation capacity is static art.
Containment doesnโt preserve advantage; it defers it to whoever builds faster.
This isnโt a job crisis. Itโs a learning-efficiency crisis (ฮทโ).
We built faster machines, not faster feedback.
โฆand now the systemโs feeling the lag.
The AI job crisis is real
The U.K. is facing a severe Gen Z job crisis: around 1.2 million graduates are competing for just 17,000 available entry-level positions, the worst ratio since the late 1990s.
AI automation and hiring freezes have wiped out many entry-level roles, while universities keep producing degrees misaligned with market needsโleaving thousands of young workers stranded between education and employment.
The U.S. is in a global race to achieve dominance in AI.
Winning the AI race will usher in a new Golden Age of human flourishing, economic competitiveness, and national securityโAmerica intends to LEAD!