AI Intent, Public Wealth
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Sunday, July 12, 2026
You spent the week telling AI what to do. Turns out you should've been telling it what you want instead. This Sunday, we've got a Fields Medalist resurrecting 27-year-old math apps in hours, Anthropic revealing that half of Claude's enterprise usage is glorified busywork, and American workers saying they want 50% of AI companies' stock handed to the public. Grab your coffee -- this one's got range.
Today in AI:
- Terry Tao Let AI Do His Homework (From 1999) - Fields Medal winner Terence Tao used a coding agent to port two dozen Java applets to modern JavaScript in hours. The agent found two bugs in his original code while introducing only one of its own. Terence Tao's Blog
- Claude's Real Job: The Work Nobody Wants - Anthropic analyzed 1.2 million Claude Cowork sessions across 600,000 organizations. Half of all usage went to status reports, onboarding checklists, and slide decks -- what Anthropic calls "the work around the work." The Decoder
- Workers Want Half of AI Companies' Stock - A Verasight survey of 1,690 adults found 69% of Americans support forcing AI firms to transfer 50% of their stock to a public sovereign wealth fund. Senator Bernie Sanders has proposed legislation to do exactly that. CNBC
- AI Demand Is "Almost Unlimited," Say the People Selling AI - Former Intel CEO Pat Gelsinger and Nebius CRO Marc Boroditsky told CNBC that demand for compute far outstrips supply. Meanwhile, Meta and xAI are renting out excess capacity, which doesn't exactly scream scarcity. CNBC
- Data Centers Meet Their Nemesis: Angry Neighbors - Community opposition to AI data centers is spreading across the U.S. and beyond, with fights over power grids, water, and land use intensifying. The movement traces back to a 2015 battle over an Apple facility in Athenry, Ireland. The Verge
- Stop Prompt Engineering, Start Intent Engineering - Security researcher Daniel Miessler argues the smartest way to use today's models is to describe outcomes, not steps. As models get smarter, your detailed instructions become a handicap, not a help. Daniel Miessler
- Phoebe Gates' AI Shopping App Caught Cookie Stuffing - Investigations by researcher Ben Edelman, Bloomberg, and Capital One Shopping found the AI shopping plugin Phia was claiming affiliate commissions on sales it didn't drive through fake clicks. Phia called it a bug. Engadget
- Australia's Copyright Fight Gets Personal - AI companies are lobbying to weaken Australia's copyright protections as PM Albanese prepares a landmark AI speech. Author Anna Funder called the scraping of her literary works a "crime" at Parliament House. The Guardian
Today's Takeaway:
Anthropic's analysis of 1.2 million Claude Cowork sessions reveals that 50% of enterprise AI usage goes to status reports, checklists, and decks -- not the ambitious automation projects companies love to pitch in earnings calls. Terence Tao's experience tells a similar story from the opposite direction: his coding agent didn't prove AI can replace mathematicians, it proved AI is spectacular at the tedious migration work humans avoid for decades. These two data points land in the same place. The highest-volume, highest-value use of AI right now isn't creative breakthroughs or strategic thinking -- it's clearing the backlog of unglamorous tasks that pile up because no one wants to own them. That's not a criticism. It's a strategy. If your team is still chasing flashy AI pilots while drowning in manual reporting, you're solving the wrong problem first. The Anthropic data and Tao's experiment both suggest the same move: point AI at the drudge work, then use the freed-up time for the things only humans do well.
"AI's killer app isn't genius work -- it's the pile of tasks everyone's been ignoring."
🔍 Myth Buster
The myth: "AI is replacing human workers and making human expertise obsolete"
The reality: Today's news reveals AI functioning as a tool that amplifies human capability rather than replacing it - the quantum AI peptide research was done by scientists using spare time and personal funds, meaning humans were still the creative and strategic drivers. Meanwhile, the 'valuemaxxing' trend shows enterprises are demanding AI deliver measurable returns alongside human decision-making, not replace the workforce wholesale. The U.S. Bureau of Labor Statistics projects that while some roles will be displaced, AI is expected to create 97 million new roles by 2025 according to the World Economic Forum, outpacing the 85 million displaced.
The nuance: Tech layoffs are genuinely surging alongside AI adoption, and Brené Brown's warning that workers are neurologically struggling with this pace of change is backed by real psychological research - the transition costs are not evenly distributed, and lower-income workers face disproportionate disruption with far fewer safety nets to absorb the shock.
The Bottom Line
The Pattern: The 600,000 organizations in Anthropic's data and one of the world's greatest mathematicians arrived at the same conclusion independently: AI's current sweet spot is the tedious maintenance layer of work that humans perpetually defer. The ambition gap between what companies market AI for and what people actually use it for is the real story of 2026.
The Other Read: You could argue this is underwhelming -- AI doing slide decks and porting old Java code isn't exactly the future anyone was promised, and it tracks with every new technology being adopted for boring stuff first. That's fair, but we favor the more interesting read: boring adoption at scale is exactly how technologies become indispensable, and 600,000 organizations building habits around daily AI use is a stronger signal than any single splashy demo.
Your Move: Open Claude, ChatGPT, or whatever you have handy this Sunday and hand it one specific task you've been putting off because it's tedious -- reformatting a document, drafting a checklist, summarizing meeting notes from last week. Time yourself. You'll have your own data point in about 15 minutes.
What We're Working On
✨ Founding Cohort Special - 60% Off! - Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now
✨ Free 30-Minute AI Consultation - Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call
✨ How AI-Fluent Are You? - Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team