Welcome, humans.

So Microsoft's AI is apparently writing code faster than developers can say “job security.”

At Meta's LlamaCon conference, CEO Satya Nadella casually dropped that 20-30% of Microsoft's code repositories are now AI-generated. The progress varies by language—Python's thriving, while C++ is still mostly humans-only territory.

Meta also apparently forecasted that its AI products could “generate” between $2-3Bin 2025, surging to as much as $1.4T by 2035.

$1.4T by 2035? That's cute—my ChatGPT just predicted I'll be a billionaire by next Tuesday... if only I'd stop asking it for dad jokes to open this email…

Here’s what you need to know about AI today:

• World unveiled its portable Orb Mini eye scanner and announced major partnerships with Visa for crypto payments and Tinder for identity verification.

• Google's CEO hopes to put Gemini AI on Apple devices by mid-2025.

• Wikipedia announced plans to use AI to assist (not replace) human volunteers.

• LM Arena was accused by Stanford of helping AI makers game benchmarks.

Sam Altman’s World eye-scanning tech goes mainstream with Visa and Tinder partnerships

Sam Altman's other big project is coming for your eyeballs—and your wallet and dating life too.

Tools for Humanity just unveiled the Orb Mini, a portable version of its iris-scanning device for the World ID human verification project. The technology, co-founded by OpenAI CEO Sam Altman, has already enrolled 26M people globally with 12M verified users.

The Orb Mini was designed with input from former Apple designer Thomas Meyerhoff, prioritizing portability while maintaining the security of its eye-scanning tech that provides a unique blockchain identifier validating human identity.

But the bigger news is World's ambitious expansion plans and major partnerships that could bring biometric identity verification into everyday activities:

• U.S. expansion: World plans to deploy 7.5K Orb devices across six U.S. cities by the end of 2025 (4x its current global footprint).

• The World Card: Partnership with Visa to launch a crypto payment card later in 2025, allowing users to spend cryptocurrencies at 150M merchants worldwide.

  • The card automatically converts crypto to fiat for merchants and offers rewards in WLD tokens.

• Tinder verification: A collaboration with Match Group to pilot identity verification on Tinder in Japan, aiming to reduce fake profiles and ensure users are connecting with real people.

• Financial integrations: World users will gain access to prediction marketplace Kalshi and can use World ID to sign in to decentralized lending platform Morpho.

World's rapid expansion and high-profile partnerships position its biometric ID system for mainstream adoption—because lets face it, we need a way to prove humanhood online.

The burning question: Will mainstream consumers be willing to have their eyes scanned for convenient payments and better dating prospects?

And what happens when biometric identity becomes a requirement rather than just an option for everyday services, like how two factor authentication (where you’re required to get a text to login) is required by many services?

Meanwhile, World's Visa partnership comes on the same day that Visa unveiled its broader “Visa Intelligent Commerce” initiative. This is Visa’s plan to open up its payment network to developers building AI shopping experiences.

The program introduces “AI-ready credit cards” (what could go wrong??) with tokenized credentials, personalization, and secure payment options with user-set guidelines—essentially preparing for a future where AI agents browse, select, and purchase products for consumers.

Visa is partnering with major AI companies including Anthropic, Microsoft, OpenAI, and others to implement this technology. Mastercard announced a similar initiative called “Mastercard Agent Pay” just a day earlier.

Google has also begun monetizing AI—through conversations. Google is now showing ads within chatbot interactions through its AdSense network.

After testing with AI startups like iAsk and Liner, Google is expanding this program as it faces increasing competition from new AI search players like Perplexity that threaten its core advertising business.

I guess the joke's on us for thinking anything would stay ad-free forever. I'm still recovering from the trauma of Instagram putting ads between every third Story. My algorithm knows I'll stop for anything with a golden retriever—and I fall for it every time!

FROM OUR PARTNERS

Scaling Security in the Age of AI: Lessons from Vanta, Wiz, & Modo Labs

Join Vanta, Wiz, and Modo Labs on May 8th for a live fireside chat where they’ll explore key questions about AI’s impact on scaling security programs–what to watch out for, how to adapt, where to adopt AI, and what to focus on next.

This interactive discussion will include:

• Insights into top-of-mind issues for GRC and security pros.

• Actionable steps to strengthen your existing program by actioning on risks.

• Key areas to focus on getting visibility into AI technologies and leveraging AI in your program.

Register to save your spot! 

Prompt Tip of the Day

Want to turn your AI writing assistant into a ruthless editor that actually improves your work? We've discovered a meta-prompt hack that transforms ChatGPT into your personal writing roast master and instant rewrite wizard.

Here's the workflow: First, ask the AI to list the top 5 weaknesses in your draft. Then, have it rewrite the piece addressing every single flaw. It's like having a snarky editor who moonlights as a ghostwriter—minus the coffee breath and passive-aggressive sticky notes.

So, this prompt is too long to share in the newsletter. You can access the full text of this prompt in our new Prompt Tips of the Day April Digest, where we’ve included every tip from the past month.

Our favorite insight? This approach isn't just about criticism. It's a stealth learning mechanism that helps you understand your writing blind spots. Each roast is a mini-masterclass in improvement.

Pro tip: The key is in the prompt. Ask for specific, constructive feedback, not just vague criticism. Think "surgical strike" over "carpet bombing" when it comes to editing.

Treats To Try.

1. *Superhuman helps you get through email 2x faster—with AI that organizes your inbox, writes emails, and reminds you to follow up. Get 1 month free of Superhuman.

2. Microsoft launched three new open AI models in their Phi family, with the top model matching OpenAI's o3-mini on math benchmarks despite being smaller—the lineup includes Phi 4 mini reasoning, reasoning, and reasoning plus; each excels at reasoning tasks and are available with permissive licenses.

3. Amazon released Nova Premier, which processes text, images, and videos with a 1M token context window—it’s good for long documents, multistep workflows, and model distillation for custom AI solutions.

4. Sans Writer offers a clean writing space free from distractions, metrics, and privacy concerns so you can focus purely on your words. —free option available, with 50% launch discount.

5. Promptaa organizes your AI prompts in one place, upgrades your basic ideas into powerful instructions, and keeps your best prompts ready for reuse..

6. KAYAK, the travel search engine that compares prices across hundreds of sites, now offers an AI tool where you can plan trips by chatting—free to try.

7. IsYourWebsiteReady checks if your website works with AI agents and tells you exactly how to fix any problems.

8. Podpod turns any article or newsletter into a podcast episode delivered directly to your podcast app—free trial, then €1.99-3.99/month.

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Around the Horn.

• LM Arena (AI benchmarker) was accused by a new paper of giving major AI companies like Meta, OpenAI, and Google unfair advantages in their benchmarking system, claiming these companies gained up to 112% performance boosts through higher sampling rates and preferential access.

• Wikipedia announced plans to use AI to assist human volunteers rather than replace them.

• Google CEO Sundar Pichai testified that the company hopes to reach an agreement with Apple by mid-2025 to include Google's Gemini AI technology on new iPhones.

• OpenAI’s roll-back of a GPT-4o update after complaints about excessive sycophancy is complete.

• AI startup Rogo, which created a chatbot that replicates investment banker functions, raised $50M in funding led by Thrive Capital.

  • XThe investment increased the four-year-old company's valuation from $80M to $350M, reflecting belief that specialized AI tools can compete with generalist models.

• Chinese AI startup DeepSeek quietly released its new Prover-V2 model on Hugging Face, only a few days after Alibaba unveiled its Qwen3 AI models.

  • The specialized math-focused model was built on DeepSeek's V3 architecture with 671 billion parameters and uses a mixture-of-experts approach.

Thursday Trivia:

A.

B.

Here are the results of last week’s poll:

Here’s what you said:

• K.J. chose A: “How can I trust AI if it can’t spell driveway ugh…sticking to team human”

• O.C. chose B: “Second floor casement window on house on right is open but there is no corresponding hole in the wall.”

• D.C. chose A: “The driveway sign, the number sign and shading on the house, and the fence linking where it overlaps the plants all have issues that break the realism. ”

A Cat's Commentary.