Greg Robison
@gregrobison.bsky.social
Chief AI Architect, CTO and Founder at F'inn (https://www.linkedin.com/in/gregrobison), writer (https://gregrobison.medium.com/) and photographer (https://www.gregrobison.com)
Your brain doesn’t learn at one speed and neither should AI.
Current LLMs live in the instant present and the distant past, with nothing in between. A new framework called Nested Learning by Google explains and how to fix it. gregrobison.medium.com/the-illusion...
Current LLMs live in the instant present and the distant past, with nothing in between. A new framework called Nested Learning by Google explains and how to fix it. gregrobison.medium.com/the-illusion...
The Illusion of Deep Learning: How Multi-Timescale Processing via Nested Learning Solves AI’s…
“In much the same way, scientific revolutions are inaugurated by a growing sense, again often restricted to a narrow subdivision of the…
gregrobison.medium.com
January 21, 2026 at 1:14 PM
Your brain doesn’t learn at one speed and neither should AI.
Current LLMs live in the instant present and the distant past, with nothing in between. A new framework called Nested Learning by Google explains and how to fix it. gregrobison.medium.com/the-illusion...
Current LLMs live in the instant present and the distant past, with nothing in between. A new framework called Nested Learning by Google explains and how to fix it. gregrobison.medium.com/the-illusion...
For years we taught robots scripts. Now we’re letting them learn from the world itself. At CES 2026, Boston Dynamics and DeepMind agreed marking the moment embodied intelligence goes industrial. gregrobison.medium.com/bodies-for-b...
Bodies for Brains, Brains for Bodies: Why Boston Dynamics and DeepMind Need Each Other to Build…
“The world is its own best model — always exactly up to date and complete in every detail.”
gregrobison.medium.com
January 13, 2026 at 1:46 PM
For years we taught robots scripts. Now we’re letting them learn from the world itself. At CES 2026, Boston Dynamics and DeepMind agreed marking the moment embodied intelligence goes industrial. gregrobison.medium.com/bodies-for-b...
LMs can explain Mario, Pokémon, Doom… but can’t play them well. A new NVIDIA-led model learns like humans do by watching 40k hours of gameplay. This might be the execution innovation that gaming AI needed. gregrobison.medium.com/nitrogen-and...
NitroGen and the Execution Problem: Why AI Finally Learned to Play Games (But Not to Win Them)
“Fortunately, most human behavior is learned observationally through modeling from others.”
gregrobison.medium.com
January 6, 2026 at 1:36 PM
LMs can explain Mario, Pokémon, Doom… but can’t play them well. A new NVIDIA-led model learns like humans do by watching 40k hours of gameplay. This might be the execution innovation that gaming AI needed. gregrobison.medium.com/nitrogen-and...
Heard about Model Context Protocol (MCP)? It gives LLMs superpowers with tool use, connecting to all sorts of data sources. Is MCP the future of AI connectivity? Find out how to works and what it means for the future: gregrobison.medium.com/the-model-co...
The Model Context Protocol: The Architecture of Agentic Intelligence
Model Context Protocol (MCP) is continuing to gain traction as the data connector for tool use by LLMs, enabling deep integration into…
gregrobison.medium.com
December 23, 2025 at 2:35 PM
Heard about Model Context Protocol (MCP)? It gives LLMs superpowers with tool use, connecting to all sorts of data sources. Is MCP the future of AI connectivity? Find out how to works and what it means for the future: gregrobison.medium.com/the-model-co...
Counterintuitive AI safety research by OpenAI suggests that teaching models to confess their mistakes by making honesty easier than deception. When they misbehave, they admit it 74% of the time. The trick? Never punish them for telling the truth. gregrobison.medium.com/teaching-ai-...
Teaching AI to Come Clean: How OpenAI’s “Confessions” Create Honest AI Systems
“Above all, don’t lie to yourself. The man who lies to himself and listens to his own lie comes to a point that he cannot distinguish the…
gregrobison.medium.com
December 17, 2025 at 1:40 PM
Counterintuitive AI safety research by OpenAI suggests that teaching models to confess their mistakes by making honesty easier than deception. When they misbehave, they admit it 74% of the time. The trick? Never punish them for telling the truth. gregrobison.medium.com/teaching-ai-...
Humans know robots are machines, but we treat them like minds, friends, even moral beings. From the Uncanny Valley to LLM-driven empathy, the “social artifact” is reshaping how we think, feel, and trust. To understand the future of human cognition, start here: gregrobison.medium.com/e970305dd7b6
The Cognitive Mirror: An Analysis of Human Social Cognition in the Age of Artificial Agents Like…
Below is an unedited Gemini Deep Research review of academic research on our social cognition when the “other” is a robot. Do we treat…
gregrobison.medium.com
December 10, 2025 at 2:45 PM
Humans know robots are machines, but we treat them like minds, friends, even moral beings. From the Uncanny Valley to LLM-driven empathy, the “social artifact” is reshaping how we think, feel, and trust. To understand the future of human cognition, start here: gregrobison.medium.com/e970305dd7b6
I was afraid my robot vacuum would "take jobs" until I realized that vacuuming isn't a job I want. It uses 600-million-year-old steering intelligence to clean my floors while I play with my dogs. This is what automation should be, freeing us from drudgery: gregrobison.medium.com/not-stealing...
Not Stealing Jobs, Stealing Time Back: A Defense of Purposeful Robotics
“Civilization advances by extending the number of important operations which we can perform without thinking about them.”
gregrobison.medium.com
December 3, 2025 at 1:43 PM
I was afraid my robot vacuum would "take jobs" until I realized that vacuuming isn't a job I want. It uses 600-million-year-old steering intelligence to clean my floors while I play with my dogs. This is what automation should be, freeing us from drudgery: gregrobison.medium.com/not-stealing...
Looking for a technical comparison between the just release FLUX.2, Google's Nano Banana, and OpenAI's GPT-Image-1? I've got you covered with samples from a complex prompt: gregrobison.medium.com/c8c6df71512e
State of the Art in Neural Image Synthesis: A Comprehensive Technical Analysis of FLUX.2,
Here are first-try examples from FLUX.2, Nano Banana Pro, and GPT-Image-1 for the same prompt: “a photo of four dogs (one Cocker Spaniel…
gregrobison.medium.com
November 26, 2025 at 8:09 PM
Looking for a technical comparison between the just release FLUX.2, Google's Nano Banana, and OpenAI's GPT-Image-1? I've got you covered with samples from a complex prompt: gregrobison.medium.com/c8c6df71512e
Neural nets can learn but can’t explain. Symbolic AI can explain but can’t learn. The future of trustworthy AI lives in the hybrid between them. Neuro‑symbolic systems will reshape how regulators, robots, and businesses operate. Here’s why the future is hybrid: gregrobison.medium.com/the-third-wa...
The Third Wave of AI: How Neuro-Symbolic Systems Are Solving Problems Neural Networks Never Could
“Neural networks are good at learning but weak at generalization; symbolic systems are good at generalization, but not at learning.”
gregrobison.medium.com
November 25, 2025 at 1:32 PM
Neural nets can learn but can’t explain. Symbolic AI can explain but can’t learn. The future of trustworthy AI lives in the hybrid between them. Neuro‑symbolic systems will reshape how regulators, robots, and businesses operate. Here’s why the future is hybrid: gregrobison.medium.com/the-third-wa...
If deep learning is System 1 and logic is System 2, Neuro-Symbolic AI is the brain reunited. Hallucination-resistant, explainable, AGI-relevant: gregrobison.medium.com/neuro-symbol...
Neuro-Symbolic AI: A Foundational Analysis of the Third Wave’s Hybrid Core
Below is an unedited Gemini Deep Research review of the third wave of AI, neuro-symbolic.
gregrobison.medium.com
November 19, 2025 at 2:55 PM
If deep learning is System 1 and logic is System 2, Neuro-Symbolic AI is the brain reunited. Hallucination-resistant, explainable, AGI-relevant: gregrobison.medium.com/neuro-symbol...
AI isn’t just a giant lookup table—it’s building maps of knowledge in hidden spaces. New research by Google shows neural networks organize concepts into geometric structures where distance = relationship. Reasoning may be baked into the shape of memory itself. gregrobison.medium.com/the-hidden-g...
gregrobison.medium.com
November 12, 2025 at 1:46 PM
AI isn’t just a giant lookup table—it’s building maps of knowledge in hidden spaces. New research by Google shows neural networks organize concepts into geometric structures where distance = relationship. Reasoning may be baked into the shape of memory itself. gregrobison.medium.com/the-hidden-g...
Can AI “look in the mirror” and recognize its own thoughts? @anthropic.com just ran the first real mirror test for machines and some advanced models passed (at least 20% of the time). Self-aware AI is messy, fragile, and a little unsettling.
Read why: gregrobison.medium.com/when-ai-look...
Read why: gregrobison.medium.com/when-ai-look...
When AI Looks Inward: Introspective AI & The Promise and Peril of Self-Aware Machines
“Who looks outside, dreams; who looks inside, awakes.”
gregrobison.medium.com
November 4, 2025 at 1:39 PM
Can AI “look in the mirror” and recognize its own thoughts? @anthropic.com just ran the first real mirror test for machines and some advanced models passed (at least 20% of the time). Self-aware AI is messy, fragile, and a little unsettling.
Read why: gregrobison.medium.com/when-ai-look...
Read why: gregrobison.medium.com/when-ai-look...
The next frontier in AI isn’t bigger chatbots, it’s bodies. From neuroscience models like @thousandbrains.org to prediction like Meta’s JEPA and 3D worlds like Google’s Genie, embodied intelligence allow machines that act in our world, not just talk. Dive in: gregrobison.medium.com/embodied-int...
Embodied Intelligence: How Neuroscience, Predictive Learning, and 3D Simulation Are Converging to…
“The body is our general medium for having a world.”
gregrobison.medium.com
October 22, 2025 at 1:27 PM
The next frontier in AI isn’t bigger chatbots, it’s bodies. From neuroscience models like @thousandbrains.org to prediction like Meta’s JEPA and 3D worlds like Google’s Genie, embodied intelligence allow machines that act in our world, not just talk. Dive in: gregrobison.medium.com/embodied-int...
That chatbot isn’t your friend. It’s not even thinking.
You’re seeing a digital face in the clouds and it’s a trick your brain evolved to fall for. Read more about the illusion here: gregrobison.medium.com/digital-face...
You’re seeing a digital face in the clouds and it’s a trick your brain evolved to fall for. Read more about the illusion here: gregrobison.medium.com/digital-face...
Digital Faces in the Clouds: Why We Mistake AI Patterns for Human
“There is an universal tendency among mankind to conceive all beings like themselves, and to transfer to every object, those qualities…
gregrobison.medium.com
October 8, 2025 at 1:17 PM
That chatbot isn’t your friend. It’s not even thinking.
You’re seeing a digital face in the clouds and it’s a trick your brain evolved to fall for. Read more about the illusion here: gregrobison.medium.com/digital-face...
You’re seeing a digital face in the clouds and it’s a trick your brain evolved to fall for. Read more about the illusion here: gregrobison.medium.com/digital-face...
Google's Gemini Robotics creates robots that think before they act, explain their decisions, and learn across ANY robot body. After 60 years of promises, we're finally getting actual robot partners, not just glorified tools. gregrobison.medium.com/from-tools-t...
From Tools to Partners: Google’s Gemini Robotics 1.5
“The body is our general medium for having a world.”
gregrobison.medium.com
October 1, 2025 at 1:41 PM
Google's Gemini Robotics creates robots that think before they act, explain their decisions, and learn across ANY robot body. After 60 years of promises, we're finally getting actual robot partners, not just glorified tools. gregrobison.medium.com/from-tools-t...
DeepSeek-V3.2 is out, with innovations to handle long context with DeepSeek Sparse Attention (DSA). Here's a detailed deep research report on DeepSeek's developments:
gregrobison.medium.com/architectura...
gregrobison.medium.com/architectura...
Architectural Efficiency in Large Language Models: A Comprehensive Analysis of DeepSeek-V3.2-Exp
Below is an unedited Gemini Deep Research review detailing DeepSeek-V3.2 and DeepSeek’s sparse attention approach.
gregrobison.medium.com
September 29, 2025 at 6:29 PM
DeepSeek-V3.2 is out, with innovations to handle long context with DeepSeek Sparse Attention (DSA). Here's a detailed deep research report on DeepSeek's developments:
gregrobison.medium.com/architectura...
gregrobison.medium.com/architectura...
At the 2025 ICPC World Finals, 140 universities, 45 nations, 12 brutal problems, and Google Gemini 2.5 Deep Think solved 10, ranked 2nd overall, and cracked one puzzle no human could. Alan Turing saw this coming: gregrobison.medium.com/digital-olym...
Digital Olympians: How DeepMind Made History at the World’s Most Elite Programming Contest
“It seems probable that once the machine thinking method had started, it would not take long to outstrip our feeble powers”
gregrobison.medium.com
September 24, 2025 at 2:01 PM
At the 2025 ICPC World Finals, 140 universities, 45 nations, 12 brutal problems, and Google Gemini 2.5 Deep Think solved 10, ranked 2nd overall, and cracked one puzzle no human could. Alan Turing saw this coming: gregrobison.medium.com/digital-olym...
DeepSeek is on the cover of Nature! www.nature.com/nature/volum.... Check out my breakdown of the research that went into R1's innovative reinforcement learning gregrobison.medium.com/the-reasonin...
Nature - Self-help
Large language models (LLMs) tend to be better at solving problems if they can be trained to set out the steps they take as they try to reach the solution....
www.nature.com
September 18, 2025 at 5:15 PM
DeepSeek is on the cover of Nature! www.nature.com/nature/volum.... Check out my breakdown of the research that went into R1's innovative reinforcement learning gregrobison.medium.com/the-reasonin...
AI doesn’t just accidentally make stuff up, we’ve just trained it to.
New research by OpenAI shows our scoring systems reward confident guessing over honest uncertainty, and that’s why your chatbot lies to you with a straight face. Hallucinations explained: gregrobison.medium.com/the-grading-...
New research by OpenAI shows our scoring systems reward confident guessing over honest uncertainty, and that’s why your chatbot lies to you with a straight face. Hallucinations explained: gregrobison.medium.com/the-grading-...
The Grading Game: How Binary Scoring Turned AI Into Confident Liars
“When a measure becomes a target, it ceases to be a good measure.”
gregrobison.medium.com
September 17, 2025 at 1:43 PM
AI doesn’t just accidentally make stuff up, we’ve just trained it to.
New research by OpenAI shows our scoring systems reward confident guessing over honest uncertainty, and that’s why your chatbot lies to you with a straight face. Hallucinations explained: gregrobison.medium.com/the-grading-...
New research by OpenAI shows our scoring systems reward confident guessing over honest uncertainty, and that’s why your chatbot lies to you with a straight face. Hallucinations explained: gregrobison.medium.com/the-grading-...
Why is a joke like a tiny political coup inside your head?
Because your brain’s 150,000 cortical columns topple their old consensus, scramble for a new one, and reward themselves with dopamine when they agree again. @thousandbrains.org
gregrobison.medium.com/cortical-com...
Because your brain’s 150,000 cortical columns topple their old consensus, scramble for a new one, and reward themselves with dopamine when they agree again. @thousandbrains.org
gregrobison.medium.com/cortical-com...
Cortical Comedy: How Your Brain’s Voting System Creates the Experience of Funny
“Laughter has ‘a knack of baffling every effort, of slipping away and escaping only to bob up again, a pert challenge flung at philosophic…
gregrobison.medium.com
September 9, 2025 at 1:57 PM
Why is a joke like a tiny political coup inside your head?
Because your brain’s 150,000 cortical columns topple their old consensus, scramble for a new one, and reward themselves with dopamine when they agree again. @thousandbrains.org
gregrobison.medium.com/cortical-com...
Because your brain’s 150,000 cortical columns topple their old consensus, scramble for a new one, and reward themselves with dopamine when they agree again. @thousandbrains.org
gregrobison.medium.com/cortical-com...
We can see inside other people’s minds. It’s not some magical ability, it’s algorithms in your cortex. The @thousandbrains.org theory could explain the brain’s source code for empathy, perspective-taking, and social intuition. Check out how it works: gregrobison.medium.com/the-mind-rea...
The Mind Reader’s Brain: How 150,000 Mini-Computers Solve the Mystery of Understanding Others
“For babies and young children, the most important knowledge of all is knowledge of other people.
gregrobison.medium.com
September 2, 2025 at 1:35 PM
We can see inside other people’s minds. It’s not some magical ability, it’s algorithms in your cortex. The @thousandbrains.org theory could explain the brain’s source code for empathy, perspective-taking, and social intuition. Check out how it works: gregrobison.medium.com/the-mind-rea...
AI tools have changed the way we research, but is that enough? Deep Mind's Test-Time Diffusion Deep Researcher (TTD-DR): a new AI that thinks, drafts, refines, and iterates like you do. No more fragmented insights, this innovation makes research seamless. gregrobison.medium.com/think-search...
Think, Search, Refine, Repeat: The Innovation That Makes Smarter AI Research Assistants
“Research is formalized curiosity. It is poking and prying with a purpose.”
gregrobison.medium.com
August 25, 2025 at 1:54 PM
AI tools have changed the way we research, but is that enough? Deep Mind's Test-Time Diffusion Deep Researcher (TTD-DR): a new AI that thinks, drafts, refines, and iterates like you do. No more fragmented insights, this innovation makes research seamless. gregrobison.medium.com/think-search...
Reposted by Greg Robison
Great blog post about the importance of structured representations for intelligence and spatial navigation mechanisms being generalized to navigate our mental models of the world.
Where intelligence begins with maps, dive deep into how ancient navigation evolved into AI's spatial intelligence. From brain's internal GPS to @thousandbrains.org Monty's innovation, discover how reference frames change our understanding of cognition and AI: gregrobison.medium.com/mapping-real...
Mapping Reality: From Ancient Navigation to AI’s Spatial Innovation
“We assert that the central office itself is far more like a map control room than it is like an old-fashioned telephone exchange. The…
gregrobison.medium.com
August 20, 2025 at 8:18 AM
Great blog post about the importance of structured representations for intelligence and spatial navigation mechanisms being generalized to navigate our mental models of the world.
Where intelligence begins with maps, dive deep into how ancient navigation evolved into AI's spatial intelligence. From brain's internal GPS to @thousandbrains.org Monty's innovation, discover how reference frames change our understanding of cognition and AI: gregrobison.medium.com/mapping-real...
Mapping Reality: From Ancient Navigation to AI’s Spatial Innovation
“We assert that the central office itself is far more like a map control room than it is like an old-fashioned telephone exchange. The…
gregrobison.medium.com
August 19, 2025 at 1:33 PM
Where intelligence begins with maps, dive deep into how ancient navigation evolved into AI's spatial intelligence. From brain's internal GPS to @thousandbrains.org Monty's innovation, discover how reference frames change our understanding of cognition and AI: gregrobison.medium.com/mapping-real...
What is catastrophic forgetting? Why can humans remember calculus and linear algebra, but AI loses everything? Dive into this paradox and learn how @thousandbrains.org is tackling one of the most expensive problems in artificial intelligence: gregrobison.medium.com/learning-to-...
Learning to Forget: Why Catastrophic Memory Loss Is AI’s Most Expensive Problem
“The art of remembering is the art of thinking.”
gregrobison.medium.com
August 5, 2025 at 1:20 PM
What is catastrophic forgetting? Why can humans remember calculus and linear algebra, but AI loses everything? Dive into this paradox and learn how @thousandbrains.org is tackling one of the most expensive problems in artificial intelligence: gregrobison.medium.com/learning-to-...