🤳 Your Selfie, But Ghibli-ified

In today’s email:

🧐 ChatGPT just turned your selfies into Studio Ghibli art, but before you dive into your Spirited Away adventure, watch out for privacy risks and legal fine print!

📚 Thinking of a PhD? Let AI be your new academic sidekick, helping with everything from summarizing papers to running experiments—just don’t skip reading the papers!

🤖 DeepSeek cracked the code to make AI actually understand humans (finally!).

Intrigued? Keep on scrolling!

🧐 Catch Behind ChatGPT’s Latest Craze

ChatGPT just broke the internet (again) — this time by letting people turn their photos into dreamy Studio Ghibli-style masterpieces. Think: you, but in Spirited Away. The internet has lost its collective mind. We’re talking 150 million users, record app activity, and one million sign-ups in an hour. Even Totoro couldn’t pull those numbers.

But before you get too excited about your anime alter ego, here are a few things to keep in mind:

⚠️ Ghibli-style fun comes with some fine print:

• 🖼️ Legal grey zone: Copying Ghibli’s vibe might not be technically illegal, but it’s toeing the copyright line.

• 🔐 Your face = data: Uploading photos = giving away biometric info. Not cute.

• 📜 Sneaky terms of service: “Irrevocable license”? Translation: they can keep and use your stuff forever.

• 🧙‍♂️ Miyazaki disapproves: The Ghibli co-founder once said AI art “disgusts” him. Yikes.

Sure, it’s magical. But there's a not-so-fun side to the Ghibli-fied gold rush—like privacy risks, unclear licensing, and the legal murkiness of mimicking iconic styles.

Have your fun, but read the fine print before you become the star of your own AI-generated fantasy adventure.

📚 AI Tools for Smarter Research

If you or someone you know is thinking about doing PhD (Pretty Huge Degree), think about letting AI into your academic life. Turns out, it’s not just for naming your cat “Sir Meows-a-Lot.” AI could quickly become your research BFF.

Take Mohammed Shafi, a civil engineering PhD student who once thought AI was just a party trick. Fast-forward to today—he’s running an AI-powered academic empire that finds papers, summarizes them, organizes citations, and even helps troubleshoot experiments.

Here’s how AI tools can level up your research game:

• Literature reviews? Tools like Deep Research & Semantic Scholar dig deep, give you citations, and save your eyeballs.

• Paper overload? “Chat with PDF” tools (like SciSpace or Claude) summarize and answer questions, so you don’t have to read everything.

• Idea drought? Visualization tools like Research Rabbit and hypothesis generators spark fresh ideas.

• Coding blues? Cursor, Claude, and GitHub Copilot help write, debug, and visualize code—without meltdowns.

• Lab support? CRESt can literally help run your experiments (yes, with robotic arms 😎).

AI can’t write your thesis (yet), but it can make the journey way less painful—and maybe even fun. Just don't forget to read your papers still... sometimes.

🧠 AI That Gets You

DeepSeek, a Chinese AI startup, just cracked a problem that’s been giving researchers a collective headache for years. They’ve figured out how to make AI understand what humans actually want—which, let’s be honest, is half the battle with chatbots these days.

They teamed up with Tsinghua University and came up with a fancy-sounding thing called “Inference-Time Scaling for Generalist Reward Modeling” (try saying that five times fast). Basically, they’ve turbo-charged how AIs get feedback from humans to become better, smarter, and way less annoying.

Here's the TL;DR in bullet-form goodness:

• 🎓 Reward models = digital teachers for AI

• 🧠 DeepSeek made them way smarter with two tricks:

  • GRM: speaks human-ish to give nuanced rewards

  • SPCT: lets the AI critique itself like a sassy student

• ⚙️ The magic happens during inference, not just training

• 💡 Smaller AIs can now punch above their weight class

They’re planning to open-source this magic soon, which could mean big things for AI systems everywhere—especially the ones that still answer “banana” to everything.

Other cool AI stuff that is trending right now 🔥🔥

😎 Google’s reportedly paying some DeepMind AI staff not to work—literally—just to keep them from jumping to rivals like OpenAI or Microsoft. - Read more

📉 AI stocks dropped after Trump imposed tariffs, raising costs for building datacenters in the U.S. - Read more

🏫 Harvard’s making all MBA students take a new AI and data science class so everyone’s ready for the future of business. - Read more

🔮 The UK’s testing out an AI that tries to predict if someone with a criminal record might turn into a future murderer — kind of like Minority Report but with spreadsheets. - Read more

🌍Europe is working to become the "AI Continent" by simplifying rules, building AI factories, and supporting startups, all while trying to balance innovation with regulation—basically, they’re aiming to beat the U.S. and China at their own AI game. - Read more