A fundamental trend in AI is that researchers conceptualize the human mind (and intelligence in general) by using as mental models the current technological tools they have available. This is implicit and insidious.
Show this thread
-
1950-1990: the dominant implicit metaphor for the mind was a computer, with formal programs and explicit knowledge bases. 2010s: the dominant implicit metaphor for the mind is now that of a freshly initialized neural network deriving knowledge and skills from "training data"
Show this thread
(both are hopelessly wrong, and it is disheartening to see the blank slate NN analogy be just as uncritically accepted by the average student today as the computer metaphor was a few decades ago)
-
-
need to add some DNA and epi-genetics ?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm glad that is not how we've been doing it at
@Numenta for the past 15 years.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Well, that's why it is critical for an AI research to become informed about cognitive psychology and biology. Mathematics and computer science isn't enough.
-
In addition, reading up on Cybernetics, Control Theory and Complex Adaptive Systems would also be helpful!
End of conversation
New conversation -
-
-
At a systems level at least, certain tasks like arithmetic and image recognition must be performed in the brain using very similar architectures to ALUs and Neural Nets. It is a reasonable *assumption* given the similarity of the output when each system is fed similar input.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Dude, it is simple. Where is the chemistry, biology, physics, and mathematics with Deep Learning? Furthermore, there’s no evolution, emotion, or qualia in Deep Learning. Finally, there’s no semblance of creativity, intuition, and meditative thought. It is pure association.
#USA1Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Interesting that you put the focus on the students and not in the system that is selling them the idea of great opportunities
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
And the pursuit of buzzwords ! Sadly research in AI is turning into a marketing race and a young-boys competition
-
And yes everyone wants to do AI, Deep Learning, Reinforcement Learning etc. Without analyzing the problem and using a method which will not be an overkill. Sometimes simple statistics and pivot tables, cross tabs and simple ML solve your problem quite well.
End of conversation
New conversation -
-
-
It’s certainly true that because we don’t understand mind we tend to think about it in metaphor. NNs have the advantage of being cognitively inspired, if nothing else.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.