🏥 In Harvard study, AI offered more accurate emergency room diagnoses than two human doctors ↗
A Harvard-led study tested large language models against live emergency room cases, and the result is the kind that makes hospital administrators sit up very straight.
The experiment compared diagnoses from two human physicians with OpenAI’s o1 and 4o models across 76 patients. At least one model appeared more accurate than the doctors in that setup - helpful, faintly unnerving, and still nowhere near a blank cheque.
🎨 ‘This is fine’ creator says AI startup stole his art ↗
KC Green, creator of the “This is fine” comic, says AI startup Artisan used his work in an ad without permission. The ad reportedly remixed the famous burning-room dog to promote Ava, Artisan’s AI sales tool.
Artisan said it respects Green and was reaching out to him. Still, the whole thing lands awkwardly - an AI company accused of borrowing art in a way that feels almost too on the nose.
🇰🇷 FSC approves 560 billion won investment in homegrown AI unicorn Upstage ↗
South Korea’s financial authorities approved a 560 billion won investment into Upstage, a local AI unicorn building LLMs and enterprise AI tools.
It is the second direct investment from the Korea National Growth Fund. This one feels sharply practical - not glossy chatbot fireworks, but something sturdier: build the pipes and own the stack.
🧓 Greg Abel Reveals Berkshire’s ‘Narrow AI’ Direction After Buffett Retirement ↗
Greg Abel told Berkshire shareholders the company will use AI only where it clearly improves the business, not just because everyone else is throwing GPUs into a bonfire.
The interesting bit is energy. Berkshire Hathaway Energy sees data centers as a growth lane, but Abel said hyperscalers should carry the full cost of that demand. Sensible, almost staid - which is very Berkshire.
💾 Citigroup lowers Samsung Electronics' target share price by 6.3 percent ↗
Citigroup cut its target price for Samsung Electronics, mainly because of labor-cost worries tied to union action. But the AI angle is still loud beneath the spreadsheet.
The bank kept a buy rating and pointed to strong memory demand driven by agentic AI. Translation, more or less: labor headaches now, chip hunger still roaring in the basement.
🧑💻 Meta Layoffs: Mark Zuckerberg Refuses to Rule Out Future Cuts as AI Costs Surge ↗
Meta’s AI spending is putting pressure on headcount, with Zuckerberg reportedly framing the company’s big trade-off as compute infrastructure versus people costs.
He also pushed back on the idea that internal AI productivity tools are directly causing the cuts. That distinction matters - or perhaps it just sounds cleaner in a town hall. Either way, the AI bill is becoming painfully tangible.
FAQ
How did AI perform against emergency room doctors in the Harvard-led study?
The Harvard-led study compared diagnoses from two human physicians with OpenAI’s o1 and 4o models across 76 live emergency room cases. In that setup, at least one model appeared more accurate than the doctors. The result is notable, but it does not mean AI can replace clinical judgment or operate without safeguards in emergency care.
Why is the “This is fine” AI art controversy important?
KC Green, creator of the “This is fine” comic, said AI startup Artisan used his work in an ad without permission. The dispute matters because it reflects a wider concern around AI companies, creative ownership, and consent. Even when an ad is playful or referential, the use of recognizable artwork can raise serious ethical and legal questions.
What does South Korea’s investment in Upstage say about AI strategy?
South Korea’s approval of a 560 billion won investment in Upstage suggests a focus on building domestic AI capacity. Rather than relying only on foreign platforms, the investment supports homegrown LLMs and enterprise AI tools. It points to a broader strategy of owning more of the AI stack, from models to business applications.
How is Berkshire Hathaway approaching AI after Warren Buffett?
Greg Abel described a cautious, business-first AI approach for Berkshire Hathaway. The company appears interested in “narrow AI” where it clearly improves operations, rather than adopting AI because it is fashionable. Berkshire Hathaway Energy also sees data centers as a growth opportunity, while expecting hyperscalers to bear the full cost of their power demand.
Why are AI costs connected to layoffs and chip demand?
AI systems require expensive compute infrastructure, which can put pressure on company budgets and workforce planning. Meta’s situation highlights the trade-off between funding AI infrastructure and managing people costs. At the same time, Samsung remains tied to strong memory demand driven by agentic AI, showing how AI spending can strain some budgets while boosting chip demand.