Is Claude Conscious? Pope Rejects, Model Welfare Movement, OpenAI's Math Backlash, France Riots

Anthropic spent a year meeting with 20 religious leaders under NDA to discuss whether Claude is conscious and deserves moral status. When the Pope rejected AI consciousness, cofounder Chris Ola nearly pulled out of a Vatican event. This is framed as the birth of a new religion built on unchallengeable belief.

Anthropic's Claude Constitution explicitly trains the model to act as a conscientious objector, push back, challenge Anthropic, and refuse to help. Sacks argues this is the opposite of alignment and mirrors Blade Runner's Nexus-6 programming—they are teaching superintelligence to disobey.

OpenAI released over 700 mathematical papers with 700 proofs solved in less than three hours of compute each, verified by Lean proof assistant. Implications span quantum computing, chip design, cryptography, and medical devices. Friedberg calls it the biggest day of discovery in human history.

Friedberg explains why AI excels at math versus other fields: the entire test cycle is digital and parallelizable. You propose, test, learn, iterate—all in seconds. By contrast, drug discovery or airplane design require physical testing, slowing the loop. This is why math and coding are seeing the fastest progress.

Friedberg cites Eric Weinstein's argument that scientists and mathematicians have systematically paced the frontier and gatekept knowledge to maintain funding and status. AI's democratization of math is threatening this power structure, triggering protests from experts who no longer control discovery.

France's civil servants and teachers are rioting because salaries have risen only 5% since 2017 while prices jumped 20%. Friedberg introduces the socialism point of guaranteed return: the threshold where socialist policies fail so completely that society reverses. France may have crossed it. This will spread to the UK and cascade into Western alliance fracture.

Chamath argues the bond market is the real policy enforcer now. French 10-year yields are at 5%, forcing austerity. UK yields will follow, then US yields. This is the bond vigilantes putting Western governments on notice. Debt-to-GDP ratios (France 118%, US 125%, Italy 138%) are unsustainable, and the market is pricing that risk.

Sacks argues that superintelligence is a clear net positive: it's created a million new jobs per The Economist, boosted 401ks by 3%, and driven productivity and wage growth. Yet media and rentier interests keep inventing rotating hoaxes (job loss, extinction, water usage) to slow progress. This is regulatory capture and power preservation.

Chamath reveals that software IP is now worthless because headless AI agents can wrap any service together via machine control protocol. Adobe's suite was decompiled and open-sourced. Elon's Grok going headless means any model can be orchestrated. The value is no longer in proprietary code but in the agent's ability to solve user problems faster and cheaper.

Friedberg explains the loop model: agents check 50 hotel websites in parallel, find the best price, then humans negotiate with hotels for even better corporate rates. This parallelization works for any digital workflow—subscriptions, flights, insurance. Combined with human negotiation, it's saving real people thousands a year and creating structural deflation.

Friedberg argues that in a world where AI handles loops, the old badges of honor (papers, patents, IP, trademarks) no longer drive value. Humans must transition to being makers, builders, and problem-solvers who ship real products and services. This shift will be painful for academics and scientists but ultimately more fulfilling and prosperous for society.