- Searle is right when he says symbols are semantically vacant (have no meaning in themselves, are not about anything per se),
- (Hence) symbols emitted by digital sensors carry with them no indication of what was sensed,
- Digital computers of the sort we have today will one day perceive, think, and learn from experience.
What has to be false in order that 1. to 3. are true?
AI has had 70 years to explain perception, generalization and inner semantics, but has only succeeded in redefining the meanings of the words “perception” and “semantics”. What can the answers to the above question say about the principles that underlie perceiving, generalizing and creating an inner semantics?
Approach: abandon the theoretical basis of AI and start again (somehow, from some basis).
Since AI has two subjects, (a) the mind, (b) the machine, abandoning AI theory means abandoning two theories:
- AI’s theory of the mind
- AI’s theory of the machine