Microsoft declares it's building "humanist superintelligence" to keep AI safe. Reality check: They're 2 years behind OpenAI, whose models they'll use until 2032. The safety pitch? Product differentiation for enterprise clients who fear runaway AI.
Three Stanford professors just raised $50M to prove OpenAI and Anthropic generate text wrong. Their diffusion models claim 10x speed by processing tokens in parallel, not sequentially. Microsoft and Nvidia are betting they're right.
Good Morning from San Francisco, OpenAI wants taxpayers to guarantee its trillion dollar dreams. Sarah Friar floated federal backstops Wednesday,
In a twist that would make economists cringe, America's latest trade policy looks like it came straight from ChatGPT. The Trump administration's new tariff calculations match what AI language models suggest when asked about trade policy.
Economists and AI researchers worldwide have noted the striking similarity to AI-generated solutions. In a New York Times podcast with Ezra Klein, Nobel laureate Paul Krugman added his analysis.
"This is more a cautionary tale about AI than economics," Krugman told Klein. The formula takes a country's trade deficit with America, divides it by imports, then cuts that number in half. Ask ChatGPT how to set tariffs, and you'll get nearly identical math.
The implications stun economists. Different countries face different tariff rates - the EU pays more than Britain. This creates an immediate problem: Companies can ship goods through low-tariff countries to dodge higher rates.
Markets hate the uncertainty. Manufacturing stocks dropped despite theoretical benefits from protectionism. Investors see what AI missed - disrupting supply chains hurts American companies that rely on international trade.
Klein pressed Krugman on why this matters. The Nobel winner explained that language models simply repeat what they find online. They can't evaluate real-world consequences or spot logical flaws.
The policy creates unprecedented confusion. Previous trade disputes followed economic logic. But when AI formulas replace human judgment, no one can predict what happens next.
Europe faces higher tariffs than Britain. Mexico pays more than Canada. Bangladesh gets hit harder than Vietnam. The numbers follow AI's math but ignore economic reality.
This marks a shift in how governments make decisions. When leaders want backup for predetermined choices, AI offers ready-made formulas that sound smart but lack wisdom.
The irony cuts deep. An administration pushing "America First" may have outsourced trade policy to artificial intelligence. The result threatens both U.S. manufacturing and international relationships.
Why this matters:
Read on, my dear:
NYT Podcast: What Trump’s Tariffs Will Actually Do | The Ezra Klein Show
OpenAI's CFO suggested federal backing for AI infrastructure at WSJ conference, as company seeks taxpayer support for $1.4 trillion buildout against $13B revenue. The ask arrives amid circular tech deals and shutdown-era austerity.
China slashes data center power bills by half—but only for domestic chips. Trump blocks Nvidia's Blackwell exports. Two governments, two subsidy strategies, one question: who can afford their industrial policy longer?
Microsoft won the first Trump-era export license to ship advanced Nvidia chips to the UAE, clearing a path for billions in data center spending. The trade: chip access for binding oversight that converts private infrastructure into alliance architecture.
OpenAI's targeting a $1 trillion IPO by 2027—the largest in history. The restructure that made it possible gave Microsoft 27% and a revenue share. Now comes the hard part: convincing public markets to fund Altman's $1.4 trillion infrastructure vision.
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.
