Part 1: Eric Schmidt and Fei-Fei Li: Human Life After Artificial Superintelligence | EP #206

Defining Superintelligence

• Eric Schmidt defines AGI as human-level intelligence and superintelligence as intelligence equal to or better than the sum of all humans.
• The 'San Francisco consensus' predicts superintelligence within 3-4 years due to compounding effects; Schmidt personally thinks it will take longer.

AI's Current Limits

• Fei-Fei Li points out AI is already superhuman in narrow areas: translating dozens of languages, fast calculation, and vast knowledge from chemistry to sports.
• She questions whether AI can ever be Newton, Einstein, or Picasso; given all celestial data, today's AI cannot deduce Newtonian laws of motion.
• Schmidt's example: a computer with all 1902 knowledge could not invent relativity. Systems cannot yet feed their own reasoning back into themselves the way mathematicians build proof on proof.

The Path Forward

• Schmidt believes real superintelligence needs another algorithmic breakthrough, citing 'non-stationarity of objectives' — systems must be able to change objectives mid-task to achieve human-like creativity.
• Brute-force reinforcement learning is combinatorially too expensive, and we are already running out of electricity.
• Diamandis notes GPT 5 Pro reportedly reached an IQ of 148, and imagines 8 billion people with an Einstein-level intelligence in their pocket via Starlink and $50 smartphones.

Economics and Abundance

• AI is projected to generate as much as 15 trillion in economic value by 2030.
• Schmidt cautions that network effects may concentrate benefits among a small number of countries, firms, and early adopters rather than distributing abundance to everyone.
• Li agrees AI democratizes healthcare, transportation, and knowledge, but says shared prosperity is a deeper social problem involving policy, politics, and distribution.

Country Strategy

• Schmidt: America leads in hyperscalers thanks to deep capital markets and TSMC chips; China is second; Saudi Arabia's strategy of partnering with America is a good example.
• Li advises countries to invest in human capital, partnerships, and their own technological stack; not investing in AI would be 'macroscopically the wrong thing to do'.
• Schmidt worries about Africa lagging without stable governments, strong universities, and industrial structures.

Discovery and World Models

• Schmidt expects the greatest near-term gains in math and software because they are scale-free and verifiable; the real world will take longer.
• Li respectfully disagrees that fundamental math and physics problems will be solved in five years, citing Einstein: most of science is asking the right question.
• World Labs, co-founded and led by Li, created the first large world model; she predicts a hybrid virtual-physical future across productivity, entertainment, communication, and education.

Human Purpose

• Schmidt: humans will still watch humans compete in sports even knowing robots could win; the win will come from teaming human judgment with supercomputers.
• Li's closing message: human dignity, human agency, and human well-being must remain at the center of AGI and ASI, whether in technology, business, product, or policy.