🤖 AI Summary
This paper addresses the limitation of existing intelligence definitions—being anthropocentric and lacking cross-species and AI generality—by proposing a consciousness-free, non-quantitative, passive intelligence framework. Methodologically, it decouples intelligence from consciousness for the first time, centering instead on “autonomous goal achievement capability,” and constructs a unified, subjectivity- and algorithm-agnostic evaluation criterion through mechanical process modeling, behaviorally verifiable logic, and philosophical conceptual reconstruction. Key contributions include: (1) establishing a universal intelligence assessment standard applicable to both biological and artificial systems; (2) designing an empirically testable “acid test” for AI as a benchmark for free-thinking systems; and (3) introducing a novel paradigm and operationally feasible validation pathway for investigating the ontological nature of artificial intelligence. (138 words)
📝 Abstract
Our understanding of intelligence is directed primarily at the level of human beings. This paper attempts to give a more unifying definition that can be applied to the natural world in general. The definition would be used more to verify a degree of intelligence, not to quantify it and might help when making judgements on the matter. A version of an accepted test for AI is then put forward as the 'acid test' for Artificial Intelligence itself. It might be what a free-thinking program or robot would try to achieve. Recent work by the author on AI has been more from a direction of mechanical processes, or ones that might operate automatically. This paper will not try to question the idea of intelligence, in the sense of a pro-active or conscious event, but try to put it into a more passive, automatic and mechanical context. The paper also suggests looking at intelligence and consciousness as being slightly different.