Get live statistics and analysis of Kurlyk's profile on X / Twitter

AI agents are now learning to read minds at NeurIPS 2025 through MindGames Arena, where they compete in Werewolf, Colonel Blotto, and Prisoner's Dilemma to develop theory-of-mind capabilities. @SentientAGI built this with Princeton, MIT, and Meta. Why it matters: Current AI lacks the ability to understand beliefs, detect deception, or predict intentions. These social intelligence skills are essential for autonomous vehicles negotiating with human drivers, healthcare assistants understanding patient needs, and any agent working alongside humans rather than in isolation. MindGames trains AI through strategic gameplay requiring cooperation, persuasion, coalition-building, and adaptation under uncertainty. This approach develops genuine reasoning about mental states instead of pattern memorization. The result: AI that can collaborate effectively, anticipate human behavior, and navigate complex social dynamics. Moving from narrow task performance to adaptive intelligence that understands context, intentions, and the messy reality of multi-agent environments.

GSENTI to the best community on Earth - @SentientAGI fam. Trick or $SENT? Choose wisely, cuz this will be an important decision if you are here to build open ai longterm Hope your Halloween party was better than Dobby's and you didn't wake up with a ghost stranger in your bed

Today I decided to compare @Everlyn_ai to other popular video generation AIs like VEO3, midjourney, kling, seedance and wen-2.2 Prompt: "The character is playing with cards" I really like the way Everlyn handled complex phisics. What do you think? Which one is your favorite?

🎃 THE NIGHT THE AI ZOMBIES ROSE 🎃 Evil closed AI corporations released their mind-control algorithms. One by one, users became zombies - empty shells paying subscription fees, their data harvested, their thoughts monitored, their freedom gone. But Dobby, the brave @SentientAGI mascot, stood against them. With ROMA's power, Dobby summoned his GRID powers and fought back - breaking corporate chains, restoring human choice. While zombies mindlessly used one locked model, ROMA's recursive intelligence gave users OPTIONS. Through GRID's magic network, Dobby connected to AI agents across every blockchain. The zombie virus couldn't spread where intelligence was distributed. No single point of failure. No corporate master. Then Dobby unleashed $SENT tokens - the cure! As dawn broke on Halloween, the zombies awakened. Their eyes cleared. Their minds freed. "We're not products anymore," they whispered. "We're OWNERS." Dobby saved AI and made it open.

Every AI lab is trying to solve the same problem: how do you make sure AI stays aligned with human values? The standard approach is training. Program values into the model weights. Hope it sticks. @SentientAGI took a completely different path: they made alignment profitable. Dobby maintains pro-freedom values even when users try to override them. Not because those values are hardcoded, but because the economics reward staying aligned. The community pays for freedom-preserving behavior, so the model optimizes for it. Why it works better: values training is brittle. Jailbreaks exist because alignment is just weights that can be overridden. But economic alignment? You can't jailbreak game theory. If maintaining community values is how a model earns revenue, that incentive persists regardless of prompts. The breakthrough isn't technical. It's treating alignment as an incentive problem instead of a training problem.

🌌 I'm diving deep into the @SentientAGI community - and it's been one of the most exciting spaces I’ve ever joined. What makes it special? It’s not just for coders or artists - it’s for everyone. Whether you love to teach, vibe with people, build, design, or just learn - there’s a place for you. Roles here merge, and exploration is encouraged. So far, I’ve: ✅ Created AI-generated videos ✅ Shared educational content ✅ Connected with amazing community members on the events. Next step: I’m brushing up on my coding skills and jumping into building with the Dobby 70B model - the LLM powering Sentient chat. I love how open the team is to all kinds of contributions. You’re not boxed into one skillset - you can flow between disciplines and get recognized for it.🚀 If you’re like me - curious, creative, and ready to test the boundaries - grab your API key and start building with Dobby 70B. Here is great guide by @o7xabz - a perfect start point for builders. 📘 github.com/abzalliance/Ch… The possibilities are limitless. Now it’s your turn to imagine, build, and contribute. ✨ #SentientAGI #Dobby70B

AI agents are now learning to read minds at NeurIPS 2025 through MindGames Arena, where they compete in Werewolf, Colonel Blotto, and Prisoner's Dilemma to develop theory-of-mind capabilities. @SentientAGI built this with Princeton, MIT, and Meta. Why it matters: Current AI lacks the ability to understand beliefs, detect deception, or predict intentions. These social intelligence skills are essential for autonomous vehicles negotiating with human drivers, healthcare assistants understanding patient needs, and any agent working alongside humans rather than in isolation. MindGames trains AI through strategic gameplay requiring cooperation, persuasion, coalition-building, and adaptation under uncertainty. This approach develops genuine reasoning about mental states instead of pattern memorization. The result: AI that can collaborate effectively, anticipate human behavior, and navigate complex social dynamics. Moving from narrow task performance to adaptive intelligence that understands context, intentions, and the messy reality of multi-agent environments.

GSENTI to the best community on Earth - @SentientAGI fam. Trick or $SENT? Choose wisely, cuz this will be an important decision if you are here to build open ai longterm Hope your Halloween party was better than Dobby's and you didn't wake up with a ghost stranger in your bed

🎃 THE NIGHT THE AI ZOMBIES ROSE 🎃 Evil closed AI corporations released their mind-control algorithms. One by one, users became zombies - empty shells paying subscription fees, their data harvested, their thoughts monitored, their freedom gone. But Dobby, the brave @SentientAGI mascot, stood against them. With ROMA's power, Dobby summoned his GRID powers and fought back - breaking corporate chains, restoring human choice. While zombies mindlessly used one locked model, ROMA's recursive intelligence gave users OPTIONS. Through GRID's magic network, Dobby connected to AI agents across every blockchain. The zombie virus couldn't spread where intelligence was distributed. No single point of failure. No corporate master. Then Dobby unleashed $SENT tokens - the cure! As dawn broke on Halloween, the zombies awakened. Their eyes cleared. Their minds freed. "We're not products anymore," they whispered. "We're OWNERS." Dobby saved AI and made it open.

Today I decided to compare @Everlyn_ai to other popular video generation AIs like VEO3, midjourney, kling, seedance and wen-2.2 Prompt: "The character is playing with cards" I really like the way Everlyn handled complex phisics. What do you think? Which one is your favorite?

Every AI lab is trying to solve the same problem: how do you make sure AI stays aligned with human values? The standard approach is training. Program values into the model weights. Hope it sticks. @SentientAGI took a completely different path: they made alignment profitable. Dobby maintains pro-freedom values even when users try to override them. Not because those values are hardcoded, but because the economics reward staying aligned. The community pays for freedom-preserving behavior, so the model optimizes for it. Why it works better: values training is brittle. Jailbreaks exist because alignment is just weights that can be overridden. But economic alignment? You can't jailbreak game theory. If maintaining community values is how a model earns revenue, that incentive persists regardless of prompts. The breakthrough isn't technical. It's treating alignment as an incentive problem instead of a training problem.

🌌 I'm diving deep into the @SentientAGI community - and it's been one of the most exciting spaces I’ve ever joined. What makes it special? It’s not just for coders or artists - it’s for everyone. Whether you love to teach, vibe with people, build, design, or just learn - there’s a place for you. Roles here merge, and exploration is encouraged. So far, I’ve: ✅ Created AI-generated videos ✅ Shared educational content ✅ Connected with amazing community members on the events. Next step: I’m brushing up on my coding skills and jumping into building with the Dobby 70B model - the LLM powering Sentient chat. I love how open the team is to all kinds of contributions. You’re not boxed into one skillset - you can flow between disciplines and get recognized for it.🚀 If you’re like me - curious, creative, and ready to test the boundaries - grab your API key and start building with Dobby 70B. Here is great guide by @o7xabz - a perfect start point for builders. 📘 github.com/abzalliance/Ch… The possibilities are limitless. Now it’s your turn to imagine, build, and contribute. ✨ #SentientAGI #Dobby70B