This ontology stuff will never work. It assumes a utopian world where people are noble and honest and consistent and smart. They are not. If something is going to work, it succeeds no matter how many bricks are thrown in the way. If it is destined to fail, no amount of money can save it. Colombian blow makes its way into the US despite all odds, you can't give away a Chevy Volt. Same with web ontology.
While this work represents obvious trajectory for programmed output AI, I think it will ultimately fail due to indirect relation btw size of KB & speed of recollection/calc. The more 'knowledge' added the longer it will take to retrieve. Also, the other downfall is introduced by inherent programmer bias, whether explicit or by method of deduction. For instance how will the KB answer a question like 'Does God exist?'. I would advise them to look at solution such as those proposed by Numenta.
Why the CYC would never be self aware and intelligent
Because at 56 min he said that children are bigger problem to understand and the research was abandoned. In reality it means the approach they have adopted is limited because WHATEVER SYSTEM CYC IS, IT SHOULD HAVE GONE THROUGH A PHASE SIMILAR TO THE ONE CHILDREN GO THROUGH IN THEIR THINKING PROCESS.REFINEMENT OF THAT PROCESS IS THE KEY TO INTELLIGENCE.
Rule: BEFORE BEING INTELLIGENT AS a ADULT WE HAVE TO BE INT AS a CHILD.
1)Imagine CYC existing 25000 years ago and reflecting the "common sense" (CS) of the humans back then. What would it be like? Would it be able to keep up with the development of human knowledge and what is considered today to be CS? Or we will always have to prime the pump?
2)Could some Cyc ontology/KB be built using sensory inputs only instead of symbolic coding?
Even if CYC fails, I think, the insights about the real world build into CYC have great value. Lenat is my hero.
This ontology stuff will never work. It assumes a utopian world where people are noble and honest and consistent and smart. They are not. If something is going to work, it succeeds no matter how many bricks are thrown in the way. If it is destined to fail, no amount of money can save it. Colombian blow makes its way into the US despite all odds, you can't give away a Chevy Volt. Same with web ontology.
oracle2world 1 month ago
While this work represents obvious trajectory for programmed output AI, I think it will ultimately fail due to indirect relation btw size of KB & speed of recollection/calc. The more 'knowledge' added the longer it will take to retrieve. Also, the other downfall is introduced by inherent programmer bias, whether explicit or by method of deduction. For instance how will the KB answer a question like 'Does God exist?'. I would advise them to look at solution such as those proposed by Numenta.
crusty7208 5 months ago
Why the CYC would never be self aware and intelligent
Because at 56 min he said that children are bigger problem to understand and the research was abandoned. In reality it means the approach they have adopted is limited because WHATEVER SYSTEM CYC IS, IT SHOULD HAVE GONE THROUGH A PHASE SIMILAR TO THE ONE CHILDREN GO THROUGH IN THEIR THINKING PROCESS.REFINEMENT OF THAT PROCESS IS THE KEY TO INTELLIGENCE.
Rule: BEFORE BEING INTELLIGENT AS a ADULT WE HAVE TO BE INT AS a CHILD.
gespilk 10 months ago in playlist AI +
1)Imagine CYC existing 25000 years ago and reflecting the "common sense" (CS) of the humans back then. What would it be like? Would it be able to keep up with the development of human knowledge and what is considered today to be CS? Or we will always have to prime the pump?
2)Could some Cyc ontology/KB be built using sensory inputs only instead of symbolic coding?
Even if CYC fails, I think, the insights about the real world build into CYC have great value. Lenat is my hero.
gespilk 1 year ago
1,000th
VIEWER!!!
PLANaceSKATEteam 3 years ago
1:01 ... a Kiwi! Hello.
thedailyenglishshow 3 years ago
18:20 "You should have brought her flowers!" lol
thedailyenglishshow 3 years ago