The revolution in AI for researchers occurred 2-3 years before the public gained access to ChatGPT in November 2022, when large neural networks trained on massive datasets began to exhibit surprising emergent properties not seen in previous AI systems.

The history of AI consists of a repeated cycle where researchers propose a new paradigm, declare it the solution to artificial intelligence, invest heavily, then abandon it when it fails to deliver; this pattern includes the General Problem Solver of the 1950s, perceptrons, expert systems, and early neural networks.

Humans learn intuitive physics (understanding that unsupported objects fall, that friction affects sliding, that pushing on a side causes rotation) through evolution and embodied experience as infants, not through language, and current AI systems lack any capacity for this kind of physical understanding.

A complete artificial intelligence system requires multiple specialized modules: a configurator (director organizing goals), perception (representing world state), a world model (predicting consequences of actions), a cost module (measuring goal satisfaction and generating emotions), and an actor (executing action sequences).

Emotions are produced by predictions about future states; if you predict a situation will result in a bad outcome and there's nothing you can do about it, that prediction generates fear, making emotions a consequence of the world model's predictions rather than immediate reactions.

To train a world model by predicting what happens in video frames doesn't work using standard LLM techniques because videos have many plausible futures and we cannot train the system to predict all possible scenarios in the high-dimensional continuous space of video frames the way we can represent discrete word probabilities.

LeCun's proposed JEPA (Joint Embedding Predictive Architecture) predicts representations of video rather than pixels, learning in abstract terms what may happen as a consequence of actions; this approach combined with world models and cost functions could enable planning and safe, controllable AI systems.

Supervised learning requires large labeled datasets where a human tells the system what each input corresponds to, making it impractical for obscure languages, unwritten languages, or specialized domains where labeled data is scarce; this limitation motivated the development of self-supervised learning.

A system trained with self-supervised learning on images needs only 2-3 examples of a new object (like an elephant) to recognize it, making it more efficient than supervised learning which requires thousands of labeled examples—this is closer to how humans learn.

The largest AI models today have several hundred billion parameters, approaching but still below the scale of human brains which have roughly 100 billion neurons with 5,000 connections each, but training models at human-scale connectivity would require tens of thousands of GPU cards that companies like Microsoft, Meta, and Google possess.

Intelligence is not inherently connected to desire to dominate; humans have domination instincts due to being hierarchical social species, while non-social species like orangutans (which are as intelligent as humans) have no desire to dominate, so creating intelligent AI systems need not create power-seeking AI.

Humans regularly work with and learn from people smarter than themselves, and future AI assistants will similarly be subservient and make humans smarter rather than dominating them, comparable to how humans managed to work with knowledge and tools more capable than individual humans.

Information about dangerous activities like making sarin gas is not available in public internet data, so LLMs trained on public data will never be able to provide instructions for biological weapons regardless of scaling, making the jailbreaking threat exaggerated.

Perfect open-source infrastructure for AI is necessary for safety because proprietary systems concentrated in a few Western companies would allow those companies to control human opinion and culture, creating regulatory capture, while no other government would accept American companies controlling super-intelligent AI systems.

Humans have evolved specialized intelligence optimized for survival in ancestral environments rather than possessing general intelligence; computers being better at chess, go, poker, and now many games shows humans are bad at these tasks, not that computers are broadly intelligent.

A 17-year-old human can learn to drive a car reliably in about 20 hours of practice, while current autonomous vehicles require thousands of hours of engineering, specialized training data, and multiple sensors, highlighting the efficiency gap between human and AI learning.

Polarization in the United States began approximately 40 years ago before the internet existed, suggesting other causes (like the abandonment of the fairness doctrine) rather than social media are responsible, and polarization has declined in countries like Germany and France which use social media equally.

Large language models that use autoregressive prediction (predicting the next word token by token) cannot perform planning because they produce one word after the other without thinking in advance about what they will say, making them purely reactive systems similar to automatic writing in poetry.

LLMs are like a tiny model of the brain's Wernicke and Broca areas (language processing regions) but lack all the other brain systems; they are not general intelligence but highly specialized language manipulation systems that cannot do the work of the rest of the brain.

Large language models can generate fluent text by predicting probability distributions over words in sequence, but this capability masks fundamental stupidity because they lack understanding of the physical world, causal reasoning, and common sense that even animals like cats possess.