Governance/sovereign/defense: Altman blueprint/Marcus critiques/halluc/media lies/Iran threats/Shield/DoD/Sarvam/China
Key Questions
What did Gary Marcus say about hallucinations in frontier LLMs?
Marcus highlighted a 4.6% hallucination rate in frontier reasoning LLMs, countering claims it's 'next to nil'.
What issues were found with Google's AI Overviews?
Testing shows Google's AI Overviews tell millions of lies per hour due to benchmark problems.
What is Sam Altman's AI blueprint?
Altman proposed taxing robots, public wealth funds, and a four-day work week as part of OpenAI's vision for the AI economy, including a New Deal decoy.
What threats from Iran involve corporate data centers?
Iran threatened US corporate AI data centers like Palantir/Maven and OpenAI projects amid Shield $12.7B and US strikes.
What is Sarvam AI's funding?
Indian AI startup Sarvam nears $350M funding at $1.5B valuation.
What are developments in China and Korea AI?
Galaxea raised $290M in China; Rebellions in Korea scaling AI with viability tests.
What risks are associated with OpenClaw and Legion?
YC highlights OpenClaw risks; Legion focuses on ethics in governance.
How is media contributing to AI misinformation?
Media spreads AI lies through stories of fakes, amplifying hallucinations and governance concerns.
Marcus on frontier halluc (4.6%)/media AI lies (Overviews millions lies/hr); Altman blueprint/robot taxes/New Deal decoy; Iran corp DCs/Palantir/Maven; Shield $12.7B/US strikes; Sarvam $1.5B; Galaxea $290M China; Rebellions Korea; Legion ethics; YC OpenClaw risks.