Discover AI Creator Profile

New Graph Diffusion Transformer #ai

Discover AI December 27, 2025

Generative AI has a dirty secret: it is terrible at graph theory. While Diffusion Models can paint masterpieces, they treat …

MiniMax M2.1 vs GLM 4.7 - TEST AI

Discover AI December 25, 2025

We have two new Ai models and I test both new models on 2 causal reasoning tests. First is my …

Mathematical Shape of AI Thoughts (Topology, Homology)

Discover AI December 24, 2025

Is it possible to grade an AI’s reasoning without actually knowing the answer? For years, we’ve assumed that to evaluate …

DeepSeek 3.2 vs MiniMax M2 (1 Sentence TEST)

Discover AI December 23, 2025

Superior Intelligence: ChatGPT 5.2 vs MiniMax M2 on my 1 sentence test. Watch my new logic test (1 sentence, no …

ChatGPT-5.2 vs Grok 4: 1 Sentence TEST

Discover AI December 23, 2025

The Intelligence of our latest AI models is just ... amazing. Since in my last video I stated the possibility, …

NeuroSymbolic T5 AI w/ World Model

Discover AI December 22, 2025

Based on the latest Google AI research T5Gemma2 I experiment to implement an adaptive physics engine into a non-GPT AI …

GPT is Not The Future of AI: NEW AI Topology

Discover AI December 21, 2025

OpenAI's IPO approaching fast, let us reconsider their technology. Given the massive limitations of GPT AI systems today, we explore …

Self Learning AI: Accelerate w/ new RL

Discover AI December 20, 2025

The frontier of LLM research has shifted decisively toward Post-Training and System 2 reasoning. We all know the recipe for …

NEW Gemini 3 FLASH vs GPT 5.2 HIGH - A Bloodbath

Discover AI December 18, 2025

NEW Gemini 3 FLASH is 4 times cheaper ($) than OpenAI's GPT-5.2 HIGH for your identical tasks. So in a …

AI is Just a Correction Term (to Physics)

Discover AI December 17, 2025

Why Your Physics Model Needs an AI Residual. Forget AGI - Sorry Sam. AI is Just a Correction Term to …

Breaking Logic Limits: NEW AI Solution Paths (GPT-5.2 high+)

Discover AI December 16, 2025

Breaking Logic Limits: NEW AI Solution Paths for your LOCAL LLM. How to optimize your local LLM with a simple …

Superhuman SPATIAL AI: Finally !

Discover AI December 15, 2025

AGI, ASI and Superhuman performance of AI: a real - world reflection on the latest AI research results for Vision …

New Paradigm: Single Layer AI

Discover AI December 14, 2025

For a decade, we believed that 'Deeper is Better.' We thought that to learn a new task or generate a …

FIRE all AI Agents! New SCALING Laws (Google, MIT)

Discover AI December 12, 2025

New research from Google DeepMind & MIT proves that the "More Agents = Better" heuristic is mathematically wrong. We analyze …

NEW GPT 5.2: A Total Bloodbath

Discover AI December 11, 2025

Brand new GPT-5.2 was released just hours ago and I tested it not on standard known benchmarks, but on my …

AI Paradox: Use Text for Logic, Avatars for Meaning

Discover AI December 11, 2025

All rights w/ authors: "Future You: Designing and Evaluating Multimodal AI-generated Digital Twins for Strengthening Future Self-Continuity" Constanze Albrecht∗ MIT …

AI Latent Space Surgery: The End of Fine-Tuning?

Discover AI December 10, 2025

Orthogonal Subspaces in AI encode ... personality alignment or more entangled complexities by placing additional attention heads on the transformer …

AI calculates Your Future Professional Career Choices (MIT)

Discover AI December 9, 2025

AI Avatars for professional choices: Simulating Your Future Professional Career Choices with AI. Can AI Predict Your Future Professional Career …

Grand Unified Theory of AI (Explained w/ Google ADK)

Discover AI December 8, 2025

Grand Unified Theory of AI Multi-Agent Systems (Explained w/ Google Context ADK). We present a rigorous analysis of Google Cloud’s …

Tokenizing Gravity Waves: AI in Astrophysics (LIGO)

Discover AI December 7, 2025

All rights W/ authors: Large Language Models for Limited Noisy Data: A Gravitational Wave Identification Study Yixuan Li ,1, 2 …

Hallucinate.DELETE = We found the H-Neurons

Discover AI December 5, 2025

NEW AI research by Tsinghua univ: For years, we have treated LLM hallucinations as an inevitable "psychological" flaw of autoregressive …

AI Swarm Intelligence for Protein Folding (MIT)

Discover AI December 4, 2025

NEW MIT AI Swarm Intelligence system for advanced Protein Folding at molecular level. All rights w/ Authors: SWARMS OF LARGE …

Reduce CONTEXT for MAX Intelligence. WHY?

Discover AI December 3, 2025

Advanced visual Ai reasoning? It is simple: create a synthetic "adversarial" environment inside the inference process of the reasoning engine. …

NEW DeepSeek V3.2 Thinking: Equal to Gemini 3 PRO?

Discover AI December 2, 2025

Finally. The improved version of DeepSeek 3.2 was released today ... With maximum Test Time Scaling. I am testing the …

Discover AI