I would absolutely disagree that we aren't "demonstrably closer to reproducing human intelligence in a machine than we were thirty years ago". This remark shows a profound lack of understanding about the different approaches to artificial intelligence and our achievements with those approaches. The two fundamental approaches to AI are through Expert Systems and Neural Networks (Genetic Algorithms being more of an Algorithm deriver then AI). In my opinion "True AI" can really only be achieved with neural networks that simulate how our own brains work. It's brain-style programming that trains a network based on a set of learning inputs and can then be used to do pattern matching based on various inputs, even inputs it wasn't trained on. This is what powers voice recognition, handwriting recognition, facial recognition, object/edge recognition (see a pattern here?), etc. The things that humans do well is where Neural Networks have excelled. In many ways I believe these "reproduce" human intelligence, and are formidable achievements. Cognition, memory, etc. are emergent properties of our own neural complexity that is still being explored and that have yet to be fully understood and given our current level of understanding are still a long ways off which is the horizon I believe you are referring to.
Technically not an expert system since it's trained on a dataset. More then likely a neural network...either Back-Propogation or possibly Kohonen. It's a nice AI innovation, but we're a long way off from ever replacing a voice. Symantics and pragmatics shape our inflection and intonation. We won't have a decent AI singing voice until we have Terminator (an AI with human-like faculties that give context to the language it learns).
I think the most interesting thing about this software is that they are obviously using some form of Neural Networks to implement the algorithm, Ok...possibly a Genetic Algorithm, but I think a NN would be more easily implemented here and more flexible because you could train more then one network with different sets of inputs and they would come up with a variety of options to choose from. Anytime a piece of software has to be "trained" it's most likely a neural network behind the scenes. Of course it won't match what a good musician could come up with anytime soon. After all a human being's set of input is not only all the music he's ever listened to, but his entire life experience.
Sure I know the difference. This is just a case of homophonic word substitution, whereas the psychological principle behind your comment...well...you are just bein an ass.
I read the first edition of GUI Bloopers: Don'ts and Do's for Software Developers and Web Designers and found it to be brilliant. It articulated many of the common sense principals I had honed as a web interface developer. As a Symbolic Systems major from Stanford I studied the psychological, physiological and neurological basis of human cognition and human computer interaction and found the book to be solidly based on many of those principals (attention span studies, visual catagorization, mental models, etc). I found it highly applicable to web design and developement, but I think the design principals are generalizable to any human computer interface really. If you are a beginner, or an expert I would still recommend this book. That being said I haven't read the second edition yet, but if you give any credence to Amazon reviews its supposed to be even better.
Re:Rule #34: Don't be the first Java site of the d
on
High Performance Web Sites
·
· Score: 4, Interesting
Rule #34 is all about how slow loading the java plugin is for any browser. It's always been slow, it was supposed to improve w/6 and really it's still slow. The main problem is that NOTHING shows up until the plugin gets loaded. My solution was two-fold. Write the object in javascript (which conveniently allows the rest of the html in the page to load and display, but also eliminates the IE problem of "click on this" to activate the applet) and create an animated gif loading screen div which I block when the applet div finished loading (I ended up loading the applet at the bottom of the page below the watermark because otherwise I couldn't catch the finished loading event. I just made the loading screen match the page background so the only way you could tell anything was going on was that the scrollbar on the right changed sizes). Not exactly elegant, but it was better then the blank screen you get waiting for the plugin to load and provided a nice custom animated loading gif instead of the default applet loading logo.
Slashdot Mirror
I would absolutely disagree that we aren't "demonstrably closer to reproducing human intelligence in a machine than we were thirty years ago". This remark shows a profound lack of understanding about the different approaches to artificial intelligence and our achievements with those approaches. The two fundamental approaches to AI are through Expert Systems and Neural Networks (Genetic Algorithms being more of an Algorithm deriver then AI). In my opinion "True AI" can really only be achieved with neural networks that simulate how our own brains work. It's brain-style programming that trains a network based on a set of learning inputs and can then be used to do pattern matching based on various inputs, even inputs it wasn't trained on. This is what powers voice recognition, handwriting recognition, facial recognition, object/edge recognition (see a pattern here?), etc. The things that humans do well is where Neural Networks have excelled. In many ways I believe these "reproduce" human intelligence, and are formidable achievements. Cognition, memory, etc. are emergent properties of our own neural complexity that is still being explored and that have yet to be fully understood and given our current level of understanding are still a long ways off which is the horizon I believe you are referring to.
Technically not an expert system since it's trained on a dataset. More then likely a neural network...either Back-Propogation or possibly Kohonen. It's a nice AI innovation, but we're a long way off from ever replacing a voice. Symantics and pragmatics shape our inflection and intonation. We won't have a decent AI singing voice until we have Terminator (an AI with human-like faculties that give context to the language it learns).
I think the most interesting thing about this software is that they are obviously using some form of Neural Networks to implement the algorithm, Ok...possibly a Genetic Algorithm, but I think a NN would be more easily implemented here and more flexible because you could train more then one network with different sets of inputs and they would come up with a variety of options to choose from. Anytime a piece of software has to be "trained" it's most likely a neural network behind the scenes. Of course it won't match what a good musician could come up with anytime soon. After all a human being's set of input is not only all the music he's ever listened to, but his entire life experience.
Sure I know the difference. This is just a case of homophonic word substitution, whereas the psychological principle behind your comment...well...you are just bein an ass.
I read the first edition of GUI Bloopers: Don'ts and Do's for Software Developers and Web Designers and found it to be brilliant. It articulated many of the common sense principals I had honed as a web interface developer. As a Symbolic Systems major from Stanford I studied the psychological, physiological and neurological basis of human cognition and human computer interaction and found the book to be solidly based on many of those principals (attention span studies, visual catagorization, mental models, etc). I found it highly applicable to web design and developement, but I think the design principals are generalizable to any human computer interface really. If you are a beginner, or an expert I would still recommend this book. That being said I haven't read the second edition yet, but if you give any credence to Amazon reviews its supposed to be even better.
-Hyperic Web Monkey
http://www.hyperic.com
Rule #34 is all about how slow loading the java plugin is for any browser. It's always been slow, it was supposed to improve w/6 and really it's still slow. The main problem is that NOTHING shows up until the plugin gets loaded. My solution was two-fold. Write the object in javascript (which conveniently allows the rest of the html in the page to load and display, but also eliminates the IE problem of "click on this" to activate the applet) and create an animated gif loading screen div which I block when the applet div finished loading (I ended up loading the applet at the bottom of the page below the watermark because otherwise I couldn't catch the finished loading event. I just made the loading screen match the page background so the only way you could tell anything was going on was that the scrollbar on the right changed sizes). Not exactly elegant, but it was better then the blank screen you get waiting for the plugin to load and provided a nice custom animated loading gif instead of the default applet loading logo.