Updated
Updated · Slate · May 25
Mathematician Calls Trillion-Parameter Chatbots Mathematical Illusions, Not Conscious Entities
Updated
Updated · Slate · May 25

Mathematician Calls Trillion-Parameter Chatbots Mathematical Illusions, Not Conscious Entities

2 articles · Updated · Slate · May 25

Summary

  • Trillion-parameter AI chatbots are still just formulas, a mathematician argues, saying systems like ChatGPT and Claude only map prompts to outputs rather than think or feel.
  • Muybridge’s 1870s horse photos provide the analogy: rapid sequences can create a convincing illusion of motion, just as rapid calculations can create a convincing illusion of consciousness.
  • Even reasoning models and multi-bot systems that appear to deliberate for minutes or interact autonomously are described as chains of discrete prompts and responses, not continuous inner experience.
  • The argument targets a widening public debate over whether current AI could become conscious, contending that inspecting the machinery resolves the question in the same way film never created living horses.

Insights

If AI consciousness is just a 'mathematical flip-book,' is our own mind just a biological one?
As AI perfectly mimics feeling, what test could ever prove it has no genuine inner world?
How do we govern AI's illusion of sentience before it reshapes our human relationships?