Page 288 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

CHAPTER

Meaning Versus

Information, Prediction 14

Versus Memory, and

Question Versus Answer

Yoonsuck Choe 1,2
1
Samsung Research, Seoul, Korea ; Department of Computer Science and Engineering,
Texas A&M University 2

CHAPTER OUTLINE
1. Introduction .......................................................................................................281
2. Meaning Versus Information................................................................................282
3. Prediction Versus Memory ..................................................................................284
4. Question Versus Answer .....................................................................................287
5. Discussion.........................................................................................................288
6. Conclusion ........................................................................................................290
Acknowledgments...................................................................................................291
References .............................................................................................................291

1. INTRODUCTION
Brain and neuroscience, psychology, artificial intelligence, all strive to understand
and replicate the functioning of the human mind. Advanced methods for imaging,
monitoring, and altering the activity of the brain at the whole-brain scale are now
available, allowing us to probe the brain in unprecedented detail. These methods
include high-resolution 3D imaging (using both physical sectioning and optical
sectioning), monitoring ongoing neural activity (calcium imaging), and altering
the activation of genetically specific neurons (optogenetics). On the other hand, in
artificial intelligence, deep learning based on decades-old neural networks research
has made exponential progress, and it is now routinely beating human performance
in many areas including object recognition and game playing.
However, despite such progress in both fields, there are still many open ques-
tions. In brain science, one of the main questions is how to put together the many
detailed experimental results into a system-level understanding of brain function.
Also, there is the ultimate question to understand the phenomenon of consciousness.
In artificial intelligence research, especially in deep learning, there are lingering
issues of robustness (for example, deep neural networks were found to be easily

281
Artificial Intelligence in the Age of Neural Networks and Brain Computing. https://doi.org/10.1016/B978-0-12-815480-9.00014-1
Copyright © 2019 Elsevier Inc. All rights reserved.