|
This week in AI
AI Stopped Assisting
This Week. It Started Acting.
AI solved an 80-year math problem, ChatGPT launched an ads platform, and Anthropic closed a round that made it the most valuable private AI company alive. Three things happened this week that, taken together, mark a turning point most people will miss.
01 · The Math Proof
An OpenAI reasoning model autonomously disproved a geometry conjecture that Paul Erdős posed in 1946. Not "AI assisted researchers." The model independently found an infinite family of constructions that beat the square grid in the unit distance problem — using algebraic number theory rather than geometry, an approach no human mathematician had attempted.
The proof was verified by Noga Alon, Melanie Wood, and Thomas Bloom — the same Bloom who publicly dismantled OpenAI's previous math claim. When your harshest critic signs off, it's not hype. It's history.
This is the first confirmed instance of AI autonomously solving a prominent open problem in mathematics. The human role was verification. Not discovery.
02 · ChatGPT Ads
OpenAI launched a self-serve Ads Manager for ChatGPT. The $50,000 minimum spend is gone. Any US business can now run CPC ad campaigns inside the AI that 400 million people use as a personal assistant. Expansion to the UK, Japan, Brazil, Mexico, and South Korea is confirmed.
The revenue targets: $2.5 billion this year. $100 billion annually by 2030. For context, Google made $307B from ads last year. OpenAI is building toward that scale — inside a tool people think of as unbiased help.
The AI's job used to be: give you the best answer. Now it's also: generate ad revenue. Those goals will eventually conflict. The best ads don't feel like ads. And AI-generated ads will feel the most natural of all.
03 · Anthropic's Valuation
Anthropic is closing a $30 billion+ funding round at a valuation above $900 billion — surpassing OpenAI's $852 billion to become the world's most valuable private AI startup. Sequoia, Dragoneer, Altimeter, and Greenoaks are each investing roughly $2 billion.
Anthropic's valuation has grown approximately 15x in 14 months. This isn't retail speculation. This is the smartest institutional capital in tech choosing the safety-first lab over the one that just launched an ad platform.
What most coverage is missing
These three stories are the same story. AI crossed from tool to independent actor (the math proof). The dominant AI product pivoted to an ad-supported model (ChatGPT ads). And the market responded by valuing the safety-first competitor higher (Anthropic's round).
"The institution capital isn't confused. It sees that trust is becoming the scarcest resource in AI — and it's pricing accordingly."
Tool of the Week
|
Perplexity Pro
AI-powered research with cited sources
In a week where AI credibility is front and center, knowing where your AI's answers come from matters. Perplexity cites its sources inline — you can verify every claim before you share it.
Why this week: The Erdős proof was verified by skeptics before anyone called it history. Apply the same standard to your own AI-assisted research. Perplexity makes that easy.
|
Prompt of the Week
Use this whenever you're evaluating whether to trust an AI tool for serious work:
|
"You are a critical evaluator of AI-generated outputs. I'm going to paste a response from an AI tool. Your job: identify every factual claim made, flag any that can't be independently verified, and rate the overall reliability of the response on a scale of 1–10. Then tell me the one question I should ask to stress-test this answer before I rely on it."
|
Works with Claude, GPT-4o, or any reasoning model. Paste any AI output after the prompt. It turns AI into its own fact-checker.
The bottom line
The week AI stopped being a tool and started acting independently was the same week the market decided that trust matters more than hype. If you're building your career on AI, pay attention to which companies are earning that trust — and which ones are monetizing yours.
|