How do we measure Artificial General Intelligence?(mb-14.github.io)
mb-14.github.io
How do we measure Artificial General Intelligence?
https://mb-14.github.io/measuring-agi
5 comments
1) Intelligence is the mitigation of uncertainty, if it does not mitigate uncertainty it is not intelligent.
2) Compose challenges in which successful criteria exists, without known or obvious established solutions. Grade these by complexity.
2) Compose challenges in which successful criteria exists, without known or obvious established solutions. Grade these by complexity.
Your first definition would include Wikipedia (or even a printed encyclopedia or other reference material, technically it would include a tape-measure, sorry.). Is Wikipedia intelligent? It is likely both Artificial and General so I think we are there, Skynet is upon us and it is user editable!
This is the difficulty of any definition of intelligence: it fails as soon as it is applied...
This is the difficulty of any definition of intelligence: it fails as soon as it is applied...
Those things you mentioned are not “mitigating” in the verb sense, the intelligence USES those things to INTELLIGENTLY mitigate “uncertainties”.
If you cannot see Wikipedia and a tape-measure as both the product and resources of intelligence, you’re failing yourself somehow.
The definition I have given remains superior!
If you cannot see Wikipedia and a tape-measure as both the product and resources of intelligence, you’re failing yourself somehow.
The definition I have given remains superior!
Btw, information is the removal of uncertainty. If it does not remove uncertainty, it is not information.
You can see how these simple truths interrelate.
You can see how these simple truths interrelate.
Then we can try to measure Intelligence.
Then we might be able to add, define and measure the other parts.
Since we cannot even define intelligence, we are a long way from measuring AGI.