Facebook's Head of AI, Yann LeCun - Should We Fear Future AI Systems?

Yann LeCun is a computer scientist with contributions in machine learning, computer vision, mobile robotics and computational neuroscience. He is well known for his work on optical character recognition and computer vision using convolutional neural networks (CNN), and is a founding father of convolutional nets. He is also one of the main creators of the DjVu image compression technology (together with Léon Bottou and Patrick Haffner). He co-developed the Lush programming language with Léon Bottou.

Recorded: October 2016

[dannygjk]

Excuse me?! His verbal legerdemain is misleading when he compares logic, inference, etc.
It is still symbol manipulation which performs functions to accomplish the same task by similar means. It is still reasonably programmable by traditional methods. Neural nets are a completely different approach.

[dannygjk]

If neural nets operated like he says then there would be no need for training. He does not seem to differentiate between training and operation of the neural net after the training phase. I have a strong feeling that this paper was written by someone else and he tried to explain the concepts. ie. he is a spokesperson.

[dannygjk]

He made a joke about most people not being able to do logic and many people in the audience started making amused sounds. That's hilarious to me because I bet most of them do poorly at logic, reasoning, and critical thinking. I have found in debates with people online that the vast majority of them would benefit by taking a philosophy course.

[dannygjk]

What we need to fear is how AI is used.

[dannygjk]

Seriously!? the trophy example would eventually be learned by a neural net system.

[dannygjk]

Um yeah..of course logic is not a synonym for knowledge. Is this guy an arts and entertainment person?