4 ms·

Thanks for reading! I'd be happy to correct specific problems if you'd be kind enough to email them to me (see the about section of my website for my email). I

Thanks for reading! I'd be happy to correct specific problems if you'd be kind enough to email them to me (see the about section of my website for my email). If not, let's stick to substantive comments on the content rather than grammar or spelling.

jcoffland 10y ago
Ok. To pick on one part. This argument is self-referencing and redundant: "However, it would be useful to have a neural network that is able to perform symbolic algebra. There are two clear reasons for this desire. First, this hypothetical system would demonstrate that neural networks can be used as a substrate for previously-achieved AI systems. Second, a neural network that could perform symbolic algebra would, by definition, be able to manipulate symbols." Then directly following the above, this sentence makes very little sense: "This would show that high level knowledge representations can be grounded in statistical models." What does it mean for "high level knowledge representations" to be "grounded in statistical models?" It sounds to me like your saying to implement symbolic algebra on NNs would prove that NNs can implement symbolic algebra. There is some good content in the article but I find the conclusions and even the premise to be overreaching. It would read better if it were condensed into a history of AI booms and busts with out making wild predictions.
- daveguy 10y ago
  The term "grounded" is to have meaning which references some external reality. He meant that a NN which can perform symbolic algebra would demonstrate that symbols are grounded in some meaning outside the identification/pattern matched sense. They have meaning outside of "an or symbol, an and symbol, a true symbol" with additional information in their relationships that a NN can understand well enough to use them in context.

4 ms·

Thanks for reading! I'd be happy to correct specific problems if you'd be kind enough to email them to me (see the about section of my website for my email). I

by kennethfriedman 10y ago

jcoffland 10y ago
Ok. To pick on one part. This argument is self-referencing and redundant: "However, it would be useful to have a neural network that is able to perform symbolic algebra. There are two clear reasons for this desire. First, this hypothetical system would demonstrate that neural networks can be used as a substrate for previously-achieved AI systems. Second, a neural network that could perform symbolic algebra would, by definition, be able to manipulate symbols." Then directly following the above, this sentence makes very little sense: "This would show that high level knowledge representations can be grounded in statistical models." What does it mean for "high level knowledge representations" to be "grounded in statistical models?" It sounds to me like your saying to implement symbolic algebra on NNs would prove that NNs can implement symbolic algebra. There is some good content in the article but I find the conclusions and even the premise to be overreaching. It would read better if it were condensed into a history of AI booms and busts with out making wild predictions.
- daveguy 10y ago
  The term "grounded" is to have meaning which references some external reality. He meant that a NN which can perform symbolic algebra would demonstrate that symbols are grounded in some meaning outside the identification/pattern matched sense. They have meaning outside of "an or symbol, an and symbol, a true symbol" with additional information in their relationships that a NN can understand well enough to use them in context.