The same rules that protect us from powerful AI could also protect its owners from competition. Chinese open models may be what keeps safety from becoming a tax on everyone else’s ability to build.

🎯 perfectly stated

Would be good to see @MikeBenzCyber lock in on this in the future. We are seeing the Google Jigsaw 2.0 origin story developing in real time. It starts w/ Moskovitz, Tallinn, EA funds, Anthropic insiders. They are pumping the extinction story, then writing the rules that lock in their lab. But what's more important is that it paves the way for the Jigsaw sequel. Google built an AI that scored political posts as “toxic,” and platforms and governments used the score to police speech. Now Anthropic and the other frontier labs build the tests that decide which models are “safe,” which researchers are credible, and which AI warnings count. Then they give those tests to Washington as the official standard. This time it's much bigger. People will outsource judgment, work, research, and memory to these systems. Whoever writes the “safety” test decides what the machine is allowed to know, say, and help you do... it will be capable of a Truman Show on steroids. Elon bought Twitter and pried open the town square. Jigsaw 2.0 could own the mind you rent next.

As intelligence becomes more abundant and capable, the limiting factor may no longer be generating insight, but translating insight into forms that human language, institutions, and shared cognition can absorb. The emerging fork in the road (critical but challenging question): If humans remain in the loop for validation, trust, governance, and coordination, language stays a hard bottleneck. If intelligence can increasingly operate through simulation, empirical feedback, and machine-to-machine coordination, then progress may continue beyond what humans can fully comprehend, and language would mainly limit our participation in it, not its existence.

AI may be the most scalable arbitrage in human history because it lets small numbers of people leverage cognition itself at near-zero marginal cost, before institutions and markets fully reprice that reality.

AI will commoditize answers faster than it commoditizes judgment. That means the value shifts away from knowing things and toward knowing which things matter, which tradeoffs govern the outcome, and what to ignore. The scarce asset is not information. It’s discernment.

Bingo!

Worth a minute or two of anyone’s time. There is so much confusion around this w/ people talking past each other using the blunt form of old school left v right or trust in govt v non trust in govt. This gets directly to the nuance & how dangerous it has become. Perfectly stated.

QE didn’t just inflate assets. It rewired morality.

Corruption isn’t tolerated. It’s rebranded.

Great interview. Worth watching.

Hot take of the day: AI made book-writing frictionless, so now everyone with an audience is cashing out via “authorship.” It’s not literature. It’s a brand carry trade. Disclaimer: All hot takes expire in 24 hours.