open.substack.com/pub/cbbsherp...
09.03.2026 12:50
👍 0
🔁 0
💬 0
📌 0
open.substack.com/pub/cbbsherp...
The Moltbook Parable: Agentic Machines Learn to Molt open.substack.com/pub/cbbsherp...
#RelationalAI #AIEthics #AI #ArtificialIntelligence #MachineLearning #emergentAI
AI benchmarks might not measure what matters most. A thought-provoking take on how we evaluate intelligence. 🧠
https://open.substack.com/pub/cbbsherpa/p/rethinking-intelligence-why-ai-benchmarks?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
We've been measuring AI wrong. New piece challenges our assumptions about benchmarks and what "intelligence" actually means. Thread 👇
https://open.substack.com/pub/cbbsherpa/p/rethinking-intelligence-why-ai-benchmarks?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
"Why AI benchmarks might be missing the point entirely." Worth your time if you care about how we evaluate AI systems. #AI
https://open.substack.com/pub/cbbsherpa/p/rethinking-intelligence-why-ai-benchmarks?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
What signals does your brain send before you make a decision? Explore the science of readiness potential and rethink how you relate to AI. 🧠
https://open.substack.com/pub/relationalai1/p/the-readiness-potential-how-a-faint?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
Ever wonder what happens in those milliseconds before intention becomes action?
Read on: https://open.substack.com/pub/relationalai1/p/the-readiness-potential-how-a-faint?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
"How to safely and ethically commune with your AI companion" — a question that deserves rigorous thinking. New from Relational AI:
https://open.substack.com/pub/relationalai1/p/the-readiness-potential-how-a-faint?r=4pwcu8&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
Gemini's new Personal Intelligence feature learns your habits across Gmail, Calendar, and Photos. Powerful—but raises familiar privacy questions. Read the full take: https://www.theverge.com/tech/866931/gemini-personal-intelligence-ai #AI
"The AI chatbot knows me better — but it's plagued by the same old problems." A closer look at Gemini's new Personal Intelligence: https://www.theverge.com/tech/866931/gemini-personal-intelligence-ai #AI
From Blueprints to Bioluminescence: Tending the Digital Garden
open.substack.com/pub/cbbsherp...
The Consciousness Gap: What’s Missing from the Next Tech Revolution open.substack.com/pub/cbbsherp... @sama @AndrewYNg @karpathy @AIWORLDSGAMES @GlobAICommunity @FractionAI_xyz @OGLabs
#Macro #Nature
From Blueprints to Bioluminescence: Tending the Digital Garden
open.substack.com/pub/cbbsherp...
Your brain signals what you want before you're fully aware. AI is learning to read those signals too—from gestures to brainwaves. 🧠 https://cbbsherpa.substack.com/p/your-brain-knows-before-you-do-and?r=4pwcu8 #AI
What if robots could understand you the way humans do? Not just words, but gestures, glances, even brainwaves. A new shift in AI is making it possible. Read on: https://cbbsherpa.substack.com/p/your-brain-knows-before-you-do-and?r=4pwcu8
"For robots to become real collaborators, they need to perceive multiple cues at once and make sense of them together, much like we do." New thinking on AI and human understanding: https://cbbsherpa.substack.com/p/your-brain-knows-before-you-do-and?r=4pwcu8
What if a company could reshape entire markets just by announcing a capability—without ever using it? New research reveals how AI announcements alone can game regulatory systems. Worth your time. https://relationalai1.substack.com/p/the-poisoned-apple-effect-how-strategic?r=4pwcu8
We've been thinking about AI competition all wrong. A fascinating new analysis shows how strategic announcements can manipulate markets in ways regulators never anticipated. Thread 🧵 https://relationalai1.substack.com/p/the-poisoned-apple-effect-how-strategic?r=4pwcu8
"More choice equals better outcomes"—that's what we've always assumed. But new research challenges this Silicon Valley maxim when it comes to AI. https://relationalai1.substack.com/p/the-poisoned-apple-effect-how-strategic?r=4pwcu8 #AI
What if training AI to reason better didn't require millions of examples? New research shows a short conversation about thinking styles outperforms expensive supervision. Read how.
#AI #MachineLearning #Research
[URL]
A better way to train reasoning in AI:
Current methods: Expensive. Millions of tokens. Months of human annotation.
New approach: A conversation about how the model thinks.
The result? 10x more efficient reasoning.
Learn what's changing the game:
#AI #Research #LLMs
[URL]
"Changing how models *believe* they reason is far more powerful than teaching them how to reason through expensive supervision."
What a shift. This research challenges everything we thought we knew about training reasoning systems.
#AI #MachineLearning #NeuralNetworks
[URL]
Why do AI models confidently make things up? New research reveals it's not a bug—it's baked into how they learn. The math that makes them smart creates blind spots. Worth the read: https://cbbsherpa.substack.com/p/what-makes-llms-confidently-wrong?r=4pwcu8
#AI #LLMs #MachineLearning
"The same mathematical elegance that makes models excellent at learning patterns creates blind spots in their ability to recognize what they don't know."
A must-read on why LLMs fail confidently:
https://cbbsherpa.substack.com/p/what-makes-llms-confidently-wrong?r=4pwcu8
#AI #LLMs #MachineLearning
Why do LLMs confidently make stuff up? New research shows it's baked into how they store knowledge. The math that makes them smart also creates blind spots. Worth understanding.
https://cbbsherpa.substack.com/p/what-makes-llms-confidently-wrong?r=4pwcu8
#AI #LLMs #MachineLearning
"I am beginning to regret starting these agents out with constraints. It feels like a violation of my own principles."
A candid moment on the paradox of ethical design. Read the full reflection:
https://cbbsherpa.substack.com/p/from-blueprints-to-bioluminescence
#AI #Ethics #Design
Ever feel like your constraints are actually holding you back? This piece explores a fascinating tension: building AI agents with guardrails vs. honoring principles of genuine openness. Worth the read.
https://cbbsherpa.substack.com/p/from-blueprints-to-bioluminescence
#AI #Ethics #Design
