- The project is based on a theory about “intelligence” as a whole (which may encompass intelligence as displayed by the human brain/mind, or may specifically refer to a class of non-human-like systems intended to display intelligence with a generality of scope at least roughly equalling that of the human brain/mind)
- There is an engineering plan to implement the above conception of intelligence in a computer system.
- Ideally, we would prefer that the project has already produced some concrete results, as publications or other structured documents, or prototypes, which can be evaluated (in the case of a proprietary project, of course, only those allowed to see the publications or prototypes may be able to do this evaluation). But this criterion is not strictly enforced here, because of the desire to give a thorough listing of relevant ideas, and the difficulty of defining it precisely.
Strictly speaking, the software systems currently existing within the AGI Projects listed here are all Proto AGI Systems rather than true AGI Systems: that is, it would appear that no one has yet succeeded in creating a software system really deserving the label General Intelligence.