Artificial Intelligence Lacks Human Thinking Ability

How Humans Anticipate Physical Actions

When a human picks up a coffee cup, predictive processing based on world understanding makes the action routine. This ability relies on constantly staying one step ahead by understanding events and their causes. For humans, grabbing a cup involves anticipating weight, grip stability, and potential outcomes like spilling or breaking. These predictions shape actions and are updated by sensory feedback, allowing continuous refinement of expectations.

AI Pattern Prediction vs. Physical Understanding

Today’s AI technologies generate text and ideas that resemble human output, drawing inspiration from artificial neural networks modeled on human neurons. However, similar results do not imply similar processes. Chatbots using large language models predict the next text fragment based on data patterns, not physical interaction. While this produces capable systems, it differs fundamentally from the physical anticipation required for tasks like robotics.

The Emergence of World Models in AI

“World models” represent an emerging AI approach aimed at helping systems learn to anticipate how situations evolve. These models can process videos, simulations, or robot recordings to predict consequences of actions. For domestic robots, this could mean anticipating that gripping too tightly might break an object or that a wrong movement could cause it to tip over, connecting predictions with physical reality through sensory information.