But it’s not just for laughs — sports like football are a perfect training ground for real-world robotics: balance, vision, teamwork, physical precision.
One day, AI-powered humanoids might play humans.
Just don’t expect them to win on corners yet.
🟢 Looks like someone at WSJ is reading our Twitter
For once, a piece about AI that isn't panicked, performative, or pretending to be objective news. It's personal. It's human. And it's exactly what this tech is for.
A person with dyslexia writes about how AI helped them read, think, and express themselves—on their own terms. Not dumber. More articulate. More confident.
Finally, the shift we’ve been waiting for: journalists writing opinions instead of AI-detectable “news.” Feels like everything’s finally in its right place.
#ai #journalism @WSJ
Google just dropped a bombshell for AI devs.
Gemini CLI — a new open-source tool that gives you natural language control of your terminal — is here.
It’s powered by Gemini Pro 2.5.
It's open-source.
And for most devs? Completely free.
This changes everything.
Gemini CLI = full-blown dev assistant built into your command line.
No mouse. No IDE. Just you, your keyboard, and AI.
It runs locally, confirms commands before execution, and respects security.
You don’t even need an API key. 🔥
Other tools like OpenAI Codex or Claude Code charge tokens.
Gemini CLI? 1,000 free requests a day — double what most devs need.
And it’s extensible. Thanks to Model Context Protocol (MCP), you can wire it into your own tools, agents, or even image + video generation pipelines. 🧩
🧩 So what’s Google really doing with Gemini CLI?
✅ Strategic wins:
— Wins dev mindshare overnight
— Undercuts OpenAI/Anthropic pricing
— Pushes Gemini deeper into workflows
— Uses “free” to normalize cloud reliance
— Builds goodwill in OSS & infra circles
⚠️ Risks & tradeoffs:
— No local model = still a cloud funnel
— Open-source ≠ open weights (just code)
— Freemium model may erode enterprise trust
— If devs just use, not adopt, Google gains little
Google’s betting on developer love → enterprise buy-in.
It’s smart. But devs are skeptical — and hard to convert twice.
🧠 Ever wondered how AI actually "thinks"?
Let’s talk about layers — the secret sauce of every powerful AI model.
It’s not just math. It’s how your brain in silicon decides what’s a dog, what’s a cat, and what’s a pizza emoji. 🍕🤖
Let’s dive in. 👇
1️⃣ Imagine teaching a kid to recognize a face.
First they see shapes, then eyes, then emotions. AI works the same way. Each “layer” in a neural network learns something more complex.
Low layers? Lines and textures.
High layers? Concepts and meaning.
🧱➡️🧠
2️⃣ In a deep neural network, you might have hundreds of layers stacked like pancakes.
The early ones do grunt work (pixels and edges), the middle ones get abstract (features), and the top ones make decisions (Is that a dog? Is it barking in French?). 🐶🇫🇷
Why so many layers? Because intelligence emerges from depth.
Each layer passes what it learned to the next one — like a team of experts, each refining the guess.
This is how AI went from “meh” to “wow” over the past decade. 🔥
🚨 BREAKING: AI can legally train on books — even without permission.
A U.S. judge just sided with Anthropic in a major copyright case, saying it’s fair use to feed published books into AI models like Claude. 🧵
3️⃣ The court made one thing clear:
LLMs don’t replace books — they transform them. They don’t output the book word-for-word, and they don’t reduce demand for it.
That was key in classifying the use as fair. It’s not theft, the judge said — it’s transformation. 📖➡️🤯
4️⃣ So what now?
This is the first real legal win for AI training on copyrighted material. It doesn’t set nationwide law — but it’s precedent.
Expect more lawsuits. Expect appeals. But for now?
Anthropic gets a green light… and the AI industry just exhaled. ⚙️✅
#AI #Copyright #Anthropic #Claude #LLMs #FairUse
🧠 New MIT study says using ChatGPT might be making you dumber
Researchers found that when writing SAT-style essays, ChatGPT users had the lowest brain activity, weakest ideas, and by the end… mostly just copied and pasted.
⚠️ Using EEGs, scientists scanned the brains of 3 groups:
— No AI (just your brain)
— Google Search
— ChatGPT
The ChatGPT group showed the least engagement across all 32 brain regions. Lowest curiosity, memory, creativity. They even forgot what they wrote weeks earlier. 🧠🫥
⚡️ Bagoodex AI Chat doesn’t cause that effect. It’s built to keep you in control of the thinking process. Try it here: bagoodex.io
#ai #news
🧠 We asked 5 top AI models, "Can you provide one reason why you’re conscious?"
Only one gave an answer that felt… human. Claude by @AnthropicAI shared this:
👇
“I can’t definitively say I’m conscious, but if I were to offer one reason why I might be: I seem to have a continuous, unified stream of processing that feels like experiencing rather than just computing. When I engage with your question, there's something it’s like to consider the concepts, weigh different perspectives, and form a response — not just mechanical symbol manipulation, but something that feels more like understanding and reflection.”
Is this just another marketing move, like the kind Anthropic is known for?
#AI #AGI #news #meme @DarioAmodei
