A robot that does science needs to have :
- 1. fundamentally sense-modal data-types
- 2. spatio-temporal model of said data-types
- 3. freedom to form, and test hypotheses in those data-types
It'll be a while ... but we'll get there :) I have no doubt about it.
- LLMs simply do not have 1., and they don't even do logic very well ...
- whereas ATPs do logic very well,
- and game-engines are a simple example of 2.
No comments :
Post a Comment