- Insane AI
- Posts
- Trump’s $500B AI bet + OpenAI’s big test
Trump’s $500B AI bet + OpenAI’s big test
ChatGPT went down today, leaving millions without their go-to AI. OpenAI confirmed “elevated error rates” are under investigation, with complaints pouring in since noon UK time. For its 300M weekly users, chaos ensued.
It’s Thursday!
ChatGPT went down today, leaving millions without their go-to AI. OpenAI confirmed “elevated error rates” are under investigation, with complaints pouring in since noon UK time. For its 300M weekly users, chaos ensued.
1: Trump unveils $500 billion AI investment amid tech feud
President Trump announced Stargate, a $500 billion AI infrastructure project spearheaded by OpenAI, Oracle, and SoftBank. The initiative aims to build data centers and energy systems in Texas, with an initial $100 billion investment. Trump called it a “declaration of confidence” in America’s AI leadership.
Tech titans clash: While Trump was joined by Sam Altman (OpenAI), Larry Ellison (Oracle), and Masayoshi Son (SoftBank), Elon Musk criticized the project on X, calling its funding claims exaggerated. Musk mocked the initiative as “hot air,” intensifying his longstanding feud with Altman, whom he once co-founded OpenAI with.
Altman shot back, defending Stargate’s scale and inviting Musk to visit its first active site. Microsoft’s Satya Nadella, committing $80 billion to Azure, added: “All I know is, I’m good for my $80B.”
Why it matters: Stargate highlights the U.S.’s push to outpace China in AI innovation, with over $1 trillion in data center investments projected. But the public spats between tech leaders underscore the challenges of collaboration in shaping AI’s future. Will partnerships like Stargate deliver results, or be derailed by rivalries?
2: When AI passes "humanity's last exam," what comes next?
AI keeps breezing through tests meant to measure its limits. Enter Humanity’s Last Exam—a grueling new evaluation designed by the Center for AI Safety and Scale AI to push AI to its intellectual edge. With 3,000 questions spanning rocket engineering to analytic philosophy, it’s billed as the hardest AI test yet.
Why it matters: AI models have surpassed benchmarks from SATs to Ph.D.-level challenges, leaving researchers struggling to measure progress. This test aims to evaluate "general intelligence"—the ability to perform complex tasks across disciplines—but raises the question: What happens when AI outpaces our ability to test it?
The test
Created by top academics, the questions underwent rigorous refinement to challenge AI while remaining solvable by humans.
Major AI models failed spectacularly—OpenAI’s top scorer managed just 8.3%—but experts believe they’ll exceed 50% within a year, signaling rapid progress.
The bigger picture
AI's development is uneven: it aces coding and diagnoses but struggles with basic tasks. The new exam highlights these capabilities while underscoring AI’s growing role in solving real-world unknowns, from scientific discoveries to economic shifts.
As AI’s capabilities advance, we may soon need new ways to measure intelligence altogether. For now, Humanity’s Last Exam offers a glimpse of the frontier we’re rapidly approaching. The question isn’t whether AI will pass—it’s what happens when testing is no longer enough.
3: The great OpenAI debate: Boom or bust?
The future of OpenAI, valued at $157 billion last October, has sparked polarizing predictions. One in four readers of The Information expect its valuation to soar past $500 billion within three years, while a nearly equal number foresee it plunging below $100 billion.
Key divides
Short-term valuation: 25% of respondents see OpenAI exceeding $500 billion in three years, while 24% predict a drop below $100 billion.
Long-term outlook: In 10 years, 40% believe OpenAI will surpass $500 billion, but 23% still expect its value to dip below $100 billion.
Public listing: Opinions are split, with 45% predicting an IPO within three years and 41% doubting it.
The context: OpenAI’s leadership is navigating governance changes that could pave the way for a public listing. Meanwhile, the race toward artificial general intelligence (AGI) adds urgency to these shifts. A majority (57%) believe OpenAI will claim AGI within three years, though others remain skeptical.
A tech turnaround: Sentiment for tech companies is brightening. While last month saw concerns about worsening conditions, 62% of respondents now expect improvement over the next six months. Meta, Amazon, and Google are among the companies riding this wave of optimism, with Meta’s recent alignment with Trump’s administration seen as a potential boost to its bottom line.
Why it matters: As OpenAI approaches critical milestones, the stakes are high for its future trajectory—and for the broader AI sector it leads. Whether it becomes a trillion-dollar titan or stumbles, OpenAI’s journey will shape the tech landscape for years to come.
More news
Editors picks ☕️
Axios: The AI personification trap.
The Information: An OpenAI math test raises eyebrows.
Business Insider: CEOs have a new skill to learn: managing AI employees.