🧠 Typing with Your Mind? It’s (Almost) Here

In today’s email:

⌨️ Forget keyboards—Meta AI’s Brain2Qwerty translates brainwaves into text. No brain surgery required!

🤔 Trusting AI with your dinner plans is easy—but what about your health? A new study reveals who’s too trusting and who’s skeptical.

🔥 The hottest AI models of 2025, from OpenAI’s budget brainiac to Google’s latest coding powerhouse.

Intrigued? Keep on scrolling!

⌨️ AI Reads Brainwaves to Type

Meta AI introduced Brain2Qwerty, an AI-powered system that translates brain signals into text—no surgery required! Unlike Elon Musk’s Neuralink (which involves, you know, brain surgery), this tech uses non-invasive methods like EEG and MEG to read your brainwaves and guess what you’re typing.

How It Works (AKA Mind-Reading 101)

• Participants typed on a keyboard while their brain activity was recorded.

• The AI then learned which brain signals matched which key presses.

• Using a mix of deep learning wizardry (transformers, convolutional networks, and a language model), Brain2Qwerty tried to reconstruct the text purely from brain signals.

The Results (Somewhat Psychic, but Not Perfect)

• MEG performed better than EEG—lower error rates and clearer signals.

• Best case? 19% error rate. Worst case? Let’s just say you wouldn’t want to send a text this way…

• The model doesn’t work in real-time yet and still needs people to type first for training.

The Bigger Picture

While Neuralink is still more accurate, it comes with brain surgery. Brain2Qwerty offers a safer, non-invasive alternative, eventually making brain-powered typing possible.

🤔 How Much Should You Trust AI?

How much do you really trust AI? Let’s be honest—letting AI pick your next Netflix binge? No problem. But letting it decide on your next surgery? Uh… maybe not.

A study from the University of South Australia found that most people are cool with AI for low-stakes stuff (music, food recs) but get skeptical when it comes to life-changing decisions (health, jobs, finances). However, people who don’t understand stats or AI much are as likely to trust an algorithm for dinner plans as they are for a medical diagnosis.

Here are some fun takeaways:

• Who’s more skeptical? Older folks, men, and people from industrialized nations (Japan, US, UK).

• Who trusts AI too much? People with low statistical literacy don’t quite get how AI works.

• What’s the issue? AI is making more and more big decisions, but we’re not keeping up with understanding how it works.

Experts say we need better AI and stats education so people know when to trust an algorithm and when to say, "Nah, I got this." AI is only as smart (or as biased) as the data it’s trained on—so let’s not blindly hand it the keys to our lives just yet.

📄 Your Cheat Sheet to the AI Boom

AI models are popping up like mushrooms after rain—except these ones might overcharge you for groceries or make up facts with confidence. Here’s a quick guide to the biggest AI drops of 2024-2025.

2025’s Freshest Models

• o3-mini (OpenAI) – The budget-friendly brainiac for coding, math, and science. Small but mighty (and cheaper than its bigger siblings).

• Deep Research (OpenAI) – Fancy AI research assistant with citations… for the price of $200/month. (College students, keep walking.)

• Le Chat (Mistral) – Fast-talking AI assistant with news smarts. Accurate? Sometimes. Entertaining? Always.

• Operator (OpenAI) – The AI intern that helps with tasks—hope it doesn’t decide you need luxury-priced eggs.

• Gemini 2.0 Pro (Google) – Super memory, great at coding, but comes with a Google One subscription fee.

2024’s AI Hall of Fame

• Sora (OpenAI) – Turns text into video, but don’t expect Hollywood-level physics.

• Llama 3.3 7B (Meta) – Open-source and efficient. A solid freebie.

• DeepSeek R1 – Stellar at coding and math, but comes with a side of Chinese government censorship.

• Grok 2 (x.AI) – Musk’s turbocharged chatbot is now faster but still limiting free users.

Other cool AI stuff that is trending right now 🔥🔥

🧩 Can AI handle brainteasers like humans, or will it throw in the towel and say, “I give up”? Researchers put top reasoning models to the test using NPR’s Sunday Puzzle—and the results are both impressive and hilariously relatable. - Read more

💔 AI is stepping in as the ultimate dating wingman—nudging guys to think twice before sending those too much, too soon messages on Tinder and Hinge. - Read more

💸 Elon Musk tried to buy OpenAI for $97.4 billion, but the board shut him down so fast they might as well have hit "decline call.” - Read more

🤖 Robots are getting a serious cash boost—Apptronik just scored $350M to make Apollo, their AI-powered humanoid, the ultimate workplace sidekick, from factory floors to space missions (and maybe even your kitchen). - Read more

🧬 What if we told you AI can now predict where proteins "hang out" in your cells—and even design new ones to go exactly where we want—potentially unlocking new ways to fight disease? - Read more