🤖 AI Backlash, Solo $10M Companies & Math's Existential Crisis

🔥 The Hank Green Cancellation: Using ChatGPT for Research Is Now Controversial

Hank Green, one of YouTube's original creators and a trusted science educator since 2007, found himself at the center of a social media firestorm over the weekend. His crime? Admitting he uses ChatGPT for research.

Green, who co-founded VidCon and runs educational channels like Crash Course, published an episode of his show Ask Hank Anything last Wednesday. In the episode, he mentioned using AI tools to compile research for questions he couldn't answer live during filming. The backlash came after his manager posted a clip where Green said "I appreciate the pushback" — a phrase some viewers flagged as stereotypical AI language.

Headlines quickly proliferated: "Hank Green Accidentally Reads AI Prompt Feedback Left in His Script." Green clarified that he hadn't used AI to write the script, but rather to pull research papers, compile quotes, and convert units — standard tasks for anyone using modern language models.

"Where did this original research come from? Pull up the paper, download the PDF, crunch it all together for me. Pull some quotes from it. Change this into different units."

Despite his explanation, the controversy didn't die down. Thousands of users criticized him for relying on AI in education, citing concerns about hallucinations and the loss of a "human element" in learning. Green, who has a published policy on AI usage at his media company, responded by saying he may reduce his publishing cadence as a result of the feedback.

The incident highlights a tension at the heart of science communication: AI is becoming deeply embedded in scientific discovery, yet vocal audiences remain skeptical or outright hostile to its use. For educators like Green, avoiding AI entirely may soon mean avoiding cutting-edge science altogether.

🧮 AI Solves Major Math Problems — But Not Everyone Is Impressed

Over the same weekend, OpenAI researcher Noam Brown announced that an internal version of Astra, OpenAI's next major model family, had solved ten major open problems in mathematics, quantum complexity, and theoretical computer science.

The achievements represent a significant leap forward in formal reasoning. Yet critics like Gary Marcus were unimpressed, tweeting: "Wake me when Astra solves a significant open-world problem that doesn't revolve around formal verification."

Others noted that the goalposts for AGI keep moving. As one commenter put it: "The goalposts are on a completely separate planet."

While it's true that AI excels in formally verifiable tasks — where success can be objectively measured — these breakthroughs remain useful. From drug efficacy to software debugging, verification matters. Still, the debate over whether this constitutes AGI or ASI remains unresolved, and those terms will likely stay vague for the foreseeable future.

😰 The Mathematician's Lament

In a viral essay published before the latest Astra news, a mathematician from the University of Auckland, Kerwin Hampshire, expressed a deep existential anxiety about AI's encroachment on his field:

"There's nothing I can do. There may be nothing you can do. I have no prescriptions, policy recommendations, or coherent call to action. I just want to be honest and open about my emotional and spiritual response. I want to feel seen."

The essay, titled The Dark Knight of Mathematics, captures the dread some academics feel as AI begins solving problems that have defined entire careers. Hampshire's concern isn't just professional — it's deeply personal. Mathematicians are trained to prove theorems, and if AI can do that faster and better, what role remains for human researchers?

The counterargument, of course, is that solving existing problems will open new questions — the entire history of science has followed this pattern. And even if AI dominates theorem-proving, teaching and application will remain human endeavors. But for those whose identity is tied to discovery, the shift feels like a coup de grâce.

💼 The Rise of the One-Person $10 Million Company

The Wall Street Journal profiled a striking new phenomenon: entrepreneurs running million-dollar businesses entirely solo, powered by AI.

Ben Broca, a 40-year-old founder, launched a company last December offering AI tools to entrepreneurs. By now, he's added 10,000 paying customers and is on track to bring in $10 million in revenue this year — all without hiring a single employee.

AI handles emails, writes and debugs code, onboards customers, and processes refunds. Broca operates from his living room in Sausalito, California, and relishes the autonomy:

"I think compromises make lukewarm results."

He's not alone. According to an analysis by Stripe, the number of solo operators on its platform generating over $1 million in revenue doubled between 2023 and 2025. The number crossing the $10 million threshold nearly tripled in the same span.

📊 How AI Is Reshaping the Labor Market

The trend is most pronounced in the information sector, where new business applications have surged nearly 45% over the past year. At the same time, the rate of applicants planning to hire workers has experienced the sharpest decline of any measured industry.

Economists see this as a strong signal that solo operators are on the rise. Julian Weisser, who runs a San Francisco-based accelerator for solo founders, put it simply:

"The bar for getting started has never been lower."

His accelerator attracted 4,500 applicants for 10 slots in its most recent cycle — nearly five times the number it drew at launch last May.

⚔️ "Everybody Has the Sword Now"

But the ease of starting a business cuts both ways. If AI makes it simple to launch, it also makes it simple to copy.

Troy Johnson, a 40-year-old entrepreneur in Orlando, runs an AI-assisted app that helps users maximize credit card benefits. He coded it entirely with AI and generates around $3,000 a month in profit with no employees.

But Johnson is acutely aware of the competitive threat:

"Everybody has the sword, and we all have the ability to unsheath Excalibur now."

The metaphor captures the double-edged nature of democratized AI tools: they lower barriers to entry, but they also intensify competition.

🤔 What Does This Mean for Jobs?

The implications for the labor market remain unclear. A recent study from Harvard Business School found that among 50,000 startups, those focused on AI operated with 25% fewer employees on average.

Yet other data suggests AI-adopting companies are hiring faster — even if they maintain lower operational headcount. The interaction of these trends is complex:

• More firms launching (because barriers are lower)
• Fewer employees per firm (because AI handles more tasks)
• Faster hiring growth (because AI accelerates scaling)

As Rembrand Koning, the Harvard professor, put it:

"If everyone's hiring less but you get four times more firms, what does that do to headcount?"

🚀 Not Everyone Succeeds

The article also profiled Samir Ahmad, a 39-year-old who left Verizon after nearly two decades to start a solo coaching business. He used AI as his "chief of staff" to develop a business plan and handle marketing.

"It was like my chief of staff, second in command."

But the business petered out within months, and Ahmad returned to a full-time corporate role at a utility company. The lesson: AI lowers the cost of starting, but it doesn't guarantee success.

💡 The Exception: Experience Still Matters

Claire Vo, a 41-year-old tech executive in San Francisco, used AI to code an app for managing product documentation. She launched it at $1 a month and within weeks had thousands of downloads.

Nearly three years later, her company has 100,000 users and is on track to make seven figures in profit this year. AI handles marketing, sales, and customer support.

But Vo cautions against over-indexing on AI's role:

"Well, AI is a shortcut. But I think people over-index on how easy AI is and under-index on how much I did to get to this point."

Her network and credibility in the industry were key. AI made execution faster, but it didn't replace the strategic thinking and relationships she'd built over years.

🎮 Is This Just a Video Game?

One open question: How many of these "businesses" are real vs. entertainment?

Platforms like Pulsia, which let users spin up AI-powered businesses, have dashboard data showing companies spending $373 on ads so far today. But the big question remains: How many are actually making more money than they spend?

As one commentator noted, Midjourney and Suno initially looked like job-killers for artists and musicians. Instead, they became toys — creative tools people enjoyed using, but not replacements for professional work.

"It's more like having a guitar you like to noodle on versus actually being a touring artist."

Could the same be true for AI-powered businesses? Possibly. But the Stripe data suggests at least some are building real, revenue-generating companies.

🏢 Meta's AI Strategy: Why They Can't Just License Like Apple

On Meta's latest earnings call, Mark Zuckerberg was asked a pointed question: Why not just license AI models from others? Apple has succeeded by partnering with labs like OpenAI. Why not follow that playbook?

Zuckerberg's answer was revealing:

"Right now, the open-source models are not as strong as the frontier models. So, no is the basic answer. Meta needs to be on the frontier with the intelligence that they use."

He also cited regulatory and policy risks. If Meta relies on Chinese open-source models, for example, they could face restrictions. And if a company offers open-source models today but pivots to closed-source tomorrow, Meta could be left stranded.

Zuckerberg also emphasized Meta's vertical integration as a core advantage:

"We're a company that builds our own data centers, our own infrastructure, our own chips, our own low-level software. When we got started, my background in engineering — I wrote a lot of the systems code. A lot of the reason why Facebook worked was because it actually just worked."

He argued that Facebook succeeded early on because it was faster and more efficient than competitors. By controlling the full stack, Meta could launch features — like Reels — that required massive compute resources. They built two extra data centers just to support the Reels recommendation algorithm, a feat that wouldn't have been possible without vertical integration.

💸 The Financial Pressure

But investors are skeptical. As Ben Thompson noted in his Stratechery analysis, Meta is "double-spending" on AI:

• Renting compute from cloud providers
• Building new data centers for future AI workloads
• Hiring researchers for MSAI (Meta's AI lab)

All of this spending comes without a clear path to monetization, and Meta's stock has suffered as a result. Zuckerberg is fighting a war on multiple fronts: capital markets, internal tension over MSAI's treatment, and the technical challenge of staying competitive with OpenAI and Google.

Still, he's not backing down.

✍️ Final Thought: Paul Graham's Book Disaster

In a lighter moment, Paul Graham shared a relatable but absurd experience on X:

"Bought a book. It was awful. Didn't want it on my shelves, but I couldn't throw it away. So it sat on a table near the door. Rushing to an appointment this morning, I grabbed it to read. First mistake. Then went to breakfast and had nothing else. So I spent the morning reading the worst book I own."

It's a perfect metaphor for 2025: sometimes you're stuck with something you didn't choose, and you just have to live with it. 📚