I know there are many cases that AI of Microsoft, Google, Facebook recognize wrong: butterfly to worm etc. I have a question about this. With current technology, can AI recognize that it is being defrauded? For simple example, a human wears a bear mask. Can AI recognize that is a human?
Hi Dang,
I like to answer a lots of questions nowadays I am seeing at RG, which is similar to this, science the AI is the only field in the word of science, which seems to knowledge and progress in that realm are grow up exponentially, and very soon may be advanced world of science to produce a new phenomena or create a some intelligent agents which are much better than humans in all aspect.of our ability and capability.(Human –[Creates]- Super Human ) So my word is the question about an AI understand that, can AI solve…., can AI identified tricky question which actually does not solvable because it does not have answered. Finally, your question is the type of task witch even human does not recognize! Because it depends on the how much the bear outfit was natural, (don’t you see a hidden camera that the guy wear the guerilla cloth and all people thought it was a real guerilla). So if we define the perfect AI intelligence level should be equal to a human, the correct action is to not recognize that it was a fake animal and just a person wearing a cloth. (I strongly suggest you read the first chapter of the Stuart Russell book – Artificial Intelligence a modern approach to read about the definition of AI in 4 type)
http://aima.cs.berkeley.edu/
*IMO the future AI agents much smarter than human and will recognize and solve such kind of problems. We will see that day soon.
Hi Dang,
I like to answer a lots of questions nowadays I am seeing at RG, which is similar to this, science the AI is the only field in the word of science, which seems to knowledge and progress in that realm are grow up exponentially, and very soon may be advanced world of science to produce a new phenomena or create a some intelligent agents which are much better than humans in all aspect.of our ability and capability.(Human –[Creates]- Super Human ) So my word is the question about an AI understand that, can AI solve…., can AI identified tricky question which actually does not solvable because it does not have answered. Finally, your question is the type of task witch even human does not recognize! Because it depends on the how much the bear outfit was natural, (don’t you see a hidden camera that the guy wear the guerilla cloth and all people thought it was a real guerilla). So if we define the perfect AI intelligence level should be equal to a human, the correct action is to not recognize that it was a fake animal and just a person wearing a cloth. (I strongly suggest you read the first chapter of the Stuart Russell book – Artificial Intelligence a modern approach to read about the definition of AI in 4 type)
http://aima.cs.berkeley.edu/
*IMO the future AI agents much smarter than human and will recognize and solve such kind of problems. We will see that day soon.
Yes and no. You can train your smartphone to recognize your face very well, also against people trying to unlock it by wearing your face-replica mask. But realistically, if your algorithm is to recognize real animal faces from realistic animal masks - it's a different story. Just remember how human faces were classified as gorrilas' faces some time ago. Definitely pattern recognition skills grow with more data and more powerful machines, but are these skills equal to "exponential progress" of machine intelligence (using the recognized knowledge in a new way beyond programmers' commands) as Amir suggests? Personally I don't see any distinct correlation yet.
@Amir: my question is not hard as you think. Human wears the bear mask (just the mask). Other parts are still of human. So everyone knows that is a human with bear mask :) The second one, I want to focus on current technology, not in future. Because in future, everything can be solved, just the time, near or far. Thank you for the docs.
@Rafal: my question is easier :) AI has not been smarter than human now yet. However, this task is easy for human (as I explain above). Thank you
This is a topic on intelligence with deception. or deception driven intelligence. Please visit my RG files for more information.
In my opinion, artificial intelligence techniques have not reached the level of human intelligence, therefore artificial intelligence techniques can not be fully trusted.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence