Toggle light / dark theme

What AI Reveals About the Brain

Can AI become smarter than humans?

In this episode, I talk to Chris Summerfield about the frontier of artificial intelligence, neuroscience, LLMs, AI agents, memory, and superintelligence.

We discuss why models like ChatGPT and Claude can feel so human, why today’s AI still does not learn like the brain, and why continual learning may be one of the most important unsolved problems in AI. Chris explains how human memory works, why sleep matters for learning, and what AI research is teaching us about intelligence itself.

We also discuss the future of work, education, creativity, and whether AI could lead to a more human world — or a much stranger one.

Topics covered:
• ⁠ ⁠Artificial intelligence and the human brain.
• ⁠ ⁠⁠LLMs, ChatGPT, Claude and AI agents.
• ⁠ ⁠⁠AI memory and continual learning.
• ⁠ AI alignment, safety and misalignment.
• ⁠. Superintelligence and self-improving systems.
• ⁠ Hallucinations, reasoning and intelligence.
• ⁠. Education, jobs and the future of work.
• ⁠. Why AI may change how humans understand themselves.

TIMESTAMPS:

With neuronal data, AI models predict grammar, meaning and context of spoken sentences

By applying machine-learning models to single-cell brain recordings taken from humans in conversation, a research team identified both individual and collective neuronal activity that reflected key features of language. The work, published in Nature, offers unprecedented insight into how neurons encode linguistic information, suggesting that brain activity may one day be used to infer speech-related thoughts, which could be transformative for some patients.

“This level of granularity is necessary for us to more completely understand how the brain generates speech and, ultimately, how we can develop technologies to restore it for individuals with communication disorders,” said Debara Tucci, M.D., director of NIH’s National Institute on Deafness and Other Communication Disorders (NIDCD).

‘The best solution is to murder him in his sleep’: AI can learn violent tendencies from each other despite zero references to violence in training data

Scientists found that AI models can inherit a taste for murder (or owls) from other models’ training data.

AI-driven optical tweezers sort hundreds of particles per hour without humans

Just as self-driving cars navigate traffic without a human behind the wheel, laboratory instruments are now being developed that can design, carry out and repeat experiments independently, 24 hours a day.

Researchers at the University of Gothenburg and other institutions have now developed an AI system capable of speeding up the operation of optical tweezers, dubbed SmartTrap. The work has been published in Nature Methods.

Silicon-compatible nanocomposite garnet enables better, simpler optical isolators

A research team from Tohoku University and Kyocera Corp. has developed a new magneto-optical material—a nanocomposite magnetic garnet film—that can be deposited directly onto silicon substrates while delivering a magneto-optical figure of merit four times higher than conventional polycrystalline films.

Using this material, the team demonstrated a monolithically integrated optical isolator on a silicon chip that matches the performance of conventional devices but with a far simpler, seed-layer-free structure. The breakthrough opens a practical path toward large-scale deployment of silicon photonics in AI-era data centers.

The work is published in the journal ACS Applied Optical Materials.

Penrose vs EWOG: Consciousness and Quantum Collapse

Consciousness beyond penrose quantum microtubules?utm_source=share&utm_medium=member_android&rcm=ACoAADcXNX8BNm6vE2wHF7V91czmcuYXcuPHhY4.


🧠⚛️ Beyond Penrose: Can Consciousness Be Derived from Geometry? For more than 30 years, Roger Penrose and Stuart Hameroff proposed that consciousness emerges through Objective Reduction (OR) inside neuronal microtubules. Penrose’s key equation is remarkably simple: τ_OR = ℏ / E_G where: τ_OR = collapse time ℏ = reduced Planck constant E_G = gravitational self-energy of the spacetime superposition The idea is: 🌌 Spacetime superposition ⟶ Gravitational instability ⟶ Wavefunction collapse ⟶ Conscious event But a major question remained: ❓ What is the mathematical mechanism that actually causes collapse? The EWOG framework attempts to provide one.

/* */