una
New Member
Posts: 12
|
Post by una on Feb 20, 2011 18:42:19 GMT -5
In the article they mentioned a Search engine and camera technology that applied common sense AI like the example of "my cat is sick".. but if someone was looking for a vet, wouldn't they automatically use their own common sense to search for "vet"? And wouldnt someone who wanted to take photos be able to decide by themselves which photos to take ? Of course its nice to have something to make life easier sometimes but wouldn't this risk making people lazy with one of the things that we. currently do better than machines, use our common sense ...?
For some decisions we rely more on emotions than common,sense, could a machine be taught to do this ?
|
|
|
Post by baldurb09 on Feb 20, 2011 19:00:01 GMT -5
(#$since ·(#$implies ··(#$and ···(#$problemDifficulty #$BabyMachine #$hard) ···(#$not ····(#$thereExistAtLeast 1 ?PERSON ·························(#$canImplement ?PERSON #$BabyMachine)))) ··(#$creatingIs #$BabyMachine #$tooHard)) ·(#$implies ··(#$and ···(#$isa ?SPECIES #$AnimalKingdon) ···(#$isLessComplex ?SPECIES ?HUMANS)) ··(#$shouldBuild ?SPECIES #$BabyMachine)))
|
|
|
Post by gudrunht on Feb 20, 2011 19:00:37 GMT -5
What I found to be of most interest is the problem with reasoning and relations between facts(knowledge) and how you would teach the difference between something that is concrete and that which is not.
If you browse the openmind webpage the problem becomes really clear. For an example, here is what the AI knows about air: (http://openmind.media.mit.edu/en/concept/air/ ) -Something you find in the air is nitrogen -Something you find in the air is birds -air is made out of nitrogen, oxygen...etc. -air is gas -air is made of molecules
So the AI has all those facts and it maps those connections/relations but how do you make it understand how nitrogen can be found in air (like birds) but that its also made out of nitrogen and that it´s also made out of molecules? And how will it distinct between a bird and nitrogen since they seem to be equal objects found in air?
|
|
|
Post by Jón Þór Kristinsson on Feb 20, 2011 19:04:59 GMT -5
Wouldn't it be simpler to start with creating a system that can use these statements toward some specific goal and then fill in the statements as needed? Instead of trying to create a database with several hundred million pieces of knowledge first then trying to find some uses for that knowledge.
|
|
|
Post by niccolo on Feb 20, 2011 22:50:50 GMT -5
I am interested in know more about these projects. What approach do you at CADIA/IIIM?
I still think that make a live-child-computer is the way. Computers do not have the common sense of 3 years kids, 3 years kids have interacted with the external world for 3*365*24h. What's the AI record of learning-interaction with the external world?
|
|