AI is impressive... but it's not human just yet.

👋 Dear Dancing Queens and Super Troupers,

Only 3 to 5 miracles stand between us and human-level AI. Yes, you read that right. Not “in a century.” Not “maybe someday.”

According to Nathan Myhrvold, former tech guru at Microsoft, artificial intelligence is just a few breakthroughs (and GPUs) away from a mind capable of understanding, reasoning, imagining… and maybe even doubting.

Meanwhile, OpenAI is sharpening GPT-5, which could launch as soon as this month — promising better reasoning, longer memory, and some serious wit. Fewer bugs, more emotion, more personality. A more human AI, once again.

And what if what truly makes us human is consciousness — that little voice in our heads constantly narrating our lives? That’s what Google is betting on, as it tries to give its robots an inner monologue…

Centaur AI, on the other hand, can literally read your mind. Or at least predict your decisions with chilling accuracy.

Can it grasp what it means to love, hesitate, or lie? Not yet. But it knows when you’re about to do it.

And if you thought art would be the last human domain standing, think again: retro-rock band The Velvet Sundown, a Spotify hit, is entirely AI-generated!

Fake faces, fake bios, fully generated music… but hundreds of thousands of plays. All presented as an “artistic provocation.”

 So — ready for the show?

👉️ AI has caught up to humans way faster than expected 😱​​​

👉️ GPT‑5 is almost ready… but you’re not ready for GPT‑5 🥴

👉️ Gemini Cricket: Google gives its robots a conscience 🤖​

👉️ Centaur knows what you’ll do… before you do 👀

👉️ This rock band doesn’t exist — but it has a million fans 🤦‍♀️​​

If we've forwarded this letter to you, subscribe by clicking on this link !

🔥 If you have 15 minutes

1️⃣AI has caught up with humans much faster than expected

The summary:  Nathan Myhrvold, former Microsoft CTO, believes artificial intelligence is already brushing up against human-level cognition.
All it would take are a few key technological breakthroughs — what he calls “miracles.” But this promising leap also brings massive ethical, energy, and existential challenges.

Details :

•  Just a few hacks short of genius: According to Myhrvold, AI is close to the goal. What remains is teaching it to reason with nuance and abstract concepts — essentially, grafting it a brain capable of intuition.

• A wild bet… that’s paying off so far: Microsoft and OpenAI have poured billions into raw computing power. Since then, AI has far outperformed their 2019 expectations. And this might just be the warm-up.

• Abstraction: the heart of intelligence: Today’s AIs excel at text generation and code tasks, but still struggle with abstract understanding. Bridging that gap will require a whole new wave of innovation.

• AI’s insatiable energy appetite: The AI boom comes with an energy explosion. Myhrvold points to next-gen reactors like TerraPower as a viable solution to the electricity dilemma. We’ll need clean, stable, scalable power to keep up.

• Poetry is cute, but utility is key: Generating haikus isn’t enough. The next phase is AI that solves real problems and transforms entire industries.

• We’re not in a Terminator movie… yet: Myhrvold chuckles at doomsday scenarios — but slips in a serious warning: a superpowered, poorly regulated AI could do worse than just crash. It could go haywire... on a global scale.

Why it's important:  AI is no longer just a lab toy. It’s about to reshape our jobs, our energy systems… and our place in the world. A few more breakthroughs, and the future will arrive — no pause button, no user manual, and with us scrambling to keep up.

2️⃣GPT‑5 is almost here… but you’re not ready for GPT‑5

The summary:  OpenAI is teasing GPT‑5, reportedly launching this July — though Sam Altman only hints at “this summer.”
Expected features include better reasoning, deeper personalization, long-context understanding, image and video generation, and smoother integration with third-party apps. In short: your tech copilot is coming soon.

Details :

•  Reasoning: No more horoscope-level rambling. GPT‑5 should improve its ability to plan and think more “like a human,” reducing hallucinations and boosting its performance in coding and summarization.

• Made-to-measure personality: Users are hoping for a more natural AI, capable of adapting its tone and style to the individual for livelier, more tailored conversations.

• XXL context window: No more frustrating limits. GPT‑5 might push context windows beyond 128,000 tokens, enabling it to handle long conversations or documents all at once.

• Image and video creation, upgraded: Expect sharper, richer visuals — and maybe even full video generation to compete with Gemini Veo 3.

• App integration on another level: GPT‑5 may plug more easily into tools like Any.do, Gmail, and more — creating a truly proactive AI.

Why it's important: If GPT‑5 lands this summer, it won’t be a minor update. It could mark the shift toward a long-term, multimodal, hyper-useful assistant. A mini-revolution in your daily life — and a serious wake-up call for the competition.

3️⃣ Gemini Cricket: Google Gives Its Robots a Conscience

The summary:  Google DeepMind has hit a new milestone: robots with an inner voice — a kind of chatty mental GPS that lets them describe what they see and do in real time.
This breakthrough helps robots learn new tasks without pre-training, by making stronger connections between observation and action. Forget clueless machines stuck in loops — these bots talk to themselves to better understand the world.

Details :

•  An internal monologue for better learning: Robots are no longer just executing commands — they’re narrating what they see and do, like keeping a robotic diary. This self-talk helps them grasp new tasks without tons of examples, like a student who gets the lesson on the first try.

• “Zero-shot” learning that stuns: Thanks to this internal voice, robots can perform completely new tasks without being trained on them beforehand. Imagine a chef nailing a recipe he’s never tried, just by thinking it through aloud — that’s game-changing efficiency.

• Less memory, more intelligence: The inner dialogue reduces the robot’s memory needs for complex tasks. Instead of being data-hungry, they become smarter at linking what they see to what they must do.

• Towards more autonomous, adaptable machines: This innovation paves the way for robots that don’t freeze at unexpected events. They “chat” internally and adapt on the fly — no more babysitting needed.

• Ethics still matter: With self-reflection comes a new question: how do we make sure their inner voices stay aligned with human values? We don’t want these bots turning into tiny prima donnas making up their own rules.

Why it's important: This is a leap toward truly autonomous robots — machines that not only perform tasks, but understand their actions.
By refining their inner voice, they reduce errors and become more reliable teammates. It’s a big step toward seamless human-robot collaboration — without the risk of being led in circles (or astray).

4️⃣Centaur Knows What You’ll Do… Before You Do

The summary:   A new AI called Centaur is making waves with its uncanny ability to predict human behavior.
By closely analyzing psychological test responses, it anticipates decisions more accurately than traditional models — like a behavioral detective powered by data instead of a GPS.

Details :

• Predictions beyond expectations: Centaur’s strength lies not just in guessing outcomes, but in decoding the mental processes behind them. It’s like reading your internal notes to forecast your next move.

• Use cases galore & a dream tool for business: This breakthrough unlocks real-world applications in targeted marketing, UX optimization, cybersecurity, and early risk behavior detection. It’s the personal coach no one asked for — but that might soon be everywhere.

• The flip side: Predicting humans also means flirting with privacy and manipulation. Ethics must act as a guardrail to keep this tech from becoming Big Brother 2.0.

• A step toward human-AI symbiosis: With Centaur, we’re entering a new era where AI better understands humans — not to overpower them, but to collaborate more smoothly.

Why it's important:  This shows AI can go beyond automation and become a true partner in decision-making.
It promises deeper insights into human behavior and more tailored interactions — but also demands serious thought on data protection and respecting individual agency.

5️⃣ This Rock Band Doesn’t Exist — But It Has a Million Fans

The summary:   A fictional band called The Velvet Sundown is making waves on Spotify with a '60s folk-rock vibe.
But here’s the twist: their music is entirely generated by AI — an artistic experiment shaking up the music industry and raising eyebrows among purists of “real sound.”

Details :

• A viral success with no human faces: The Velvet Sundown quickly won over listeners on Spotify with their debut album Floating on Echoes, released on June 5, 2025. Their single Dust on the Wind topped the Viral 50 charts in several countries, racking up over a million monthly streams — all without ever touring.

• Authenticity in question: Despite the hype, suspicions grew. No band members in sight, visuals a bit too perfect… Was it all robots in disguise? Spoiler: pretty much.

• Mystery revealed: music with no musicians: On July 5, the band confirmed it: all their music is AI-generated, using tools like Suno to blend folk melodies and synthetic vocals — guided by a human creative team.

• Mixed reactions from listeners: Some applaud the emotional quality of the tracks, while others fear a future where artificial creations replace real artists. The community is now split in a heated debate.

• Ethical and commercial stakes: The case reignites concerns about copyright and AI regulation. Industry figures and organizations are calling for clear rules to protect original creators in this new landscape.

Why it's important:  This story shows how AI is crashing the gates of the music industry — with serious skill and serious questions.
Who’s the real composer? Who’s the creator? And most of all… will we still need musicians, or just a power plug?

❤️ Tool of the Week: HeyGen Video Agent — Pro-Grade Video in One Prompt

The HeyGen Video Agent | World's First Creative Operating System

With HeyGen Video Agent, you no longer need to be an editor, scriptwriter, or voice actor: just type an idea, and it generates a ready-to-publish video. Script, visuals, voice-over, transitions — all fully automated.
It’s like ChatGPT, Adobe Premiere, and a TikTok influencer fused into your browser.

What it’s for :

•  Turn a prompt into a complete video: Say what you want (e.g., “pitch my app like a Marvel trailer”) — and it handles everything.

• Automatic scriptwriting, tailored to your tone (fun, pro, educational…).

• Image and visual generation, b-roll included.

• Add a realistic voice-over, with multilingual support and natural intonation — ready for global distribution.

• Professional editing with subtitles and transitions — without touching a single software tool.

How to use it:
Via the HeyGen interface (Pro or Agency plan). The tool is still in early access but already impressive.
Want in? Join the waitlist here.