Are AI models conscious? Do Claude and ChatGPT have an inner sentient life? At least three leading giant AI companies—Anthropic, Meta, and Alphabet—are taking those questions extremely seriously. The Financial Times reported last week that they have hired philosophers, psychologists, and ethicists to test their models’ sentience.
Geoffrey Hinton, a Nobel laureate in physics and one of the fathers of neural net research, believes AI already is conscious. Jack Clark, co-founder of Anthropic, thinks the AIs might be conscious. As the models become more sophisticated, these believers are likely to become even more certain. On social media, I commonly see comments, including from very educated and sophisticated observers, that the most advanced AI models have internal life experiences of some kind.
I am here to tell you that there is no ghost in the machine. But perhaps more importantly, there is barely a “ghost” in your own human machine. “Are people conscious?” is a better and more scientifically plausible question than whether AIs are conscious.
If there is one near-universal tendency of humans, it is to attribute intent where none is present. Prehistoric humans anthropomorphized nature and attributed natural events to good and bad deities. These kinds of beliefs persist today, not only in the folk religions of the world, but in human obsessions with fortune tellers, tarot cards, and the supernatural.
