Monday, October 24, 2011

A Lot of Answers to A Lot of Questions

Hey, thanks for all the comments! You asked a lot of questions, so I am going to try my best to give a lot of answers. The most popular question, seeing as though anonymous asked it twice, was about how you can make money off of AI. There's a lot of ways, and most of them are in fact because of all of this new technology, anonymous. Take this article for example:


Now, as you can see by who published it, it is not an article for a technological news, but rather for business news. Business loves this stuff. And look what it's about! Facial recognition! Which, like natural language processing (Siri), is a big field of study in AI, and is going to be a daily part of our lives soon if these recent applications of them in smartphones end up working out (I personally can imagine myself using Siri for anything beyond silly hi-jinks: http://sh*tsirisays.com/tagged/popular, but people seem to like her). So people who can apply this new research in AI to new technology are in demand and will remain in demand as these technologies evolve. Anything that makes life easier for people is a good thing to make money off of. And Paige, I think AI's fun, and also think that anything fun should be considered beneficial to mankind, so I think its funding is beneficial regardless of how the economies doing.

The other questioned that was asked more than once was about the evolution of dreams. Though not exactly a topic in AI, I studied it in Psych 475, which is history and systems in psychology, I believe. At the end of the semester you and a group pick one broad topic of study in psychology and each report on different aspects of it, and my group did evolutionary psychology. My topic was, being a cog sci student, the evolutionary function of consciousness, and dreams was one of the areas that was talked about a lot. As for useful functions, I'm unsure, but my favorite theory was proposed by Baumeister and Masicampo in this paper:


Most of the theories in this paper revolve around the function of simulation in a social world, so they see dreams as a way of simulating and practicing social interactions. It is important to note, however, that most things written about his topic, and similar topics, are mostly theoretical, and have to be taken with caution and an open mind. I try to remain a little skeptical on what they say because of this, but I personally like how they give different aspects of the mind applicable functions. I have had an interest in studying some of the theories they propose further, but have yet to get a chance, so maybe one of you could do it so I don't have to. (I should also probably note that this is a long paper and I did not reread it before posting it, so I sure hope they talk about dreams in somewhere in there.) And as for Tara O's question about dreaming in AI, I have never looked into it, so haven't come across any research about it, but I'm sure someone has published some. Look into it and relay your findings to me.

Now Justin, I'm going to be completely honest with you, I would never recommend anything I have read about natural language processing to anyone in the world. I think its a great thing to think about and study, but the papers I have personally read on it are some of the driest out there. If you go on to study NLP, which I would encourage, and find a paper that is not only comprehensive of whats happening the field, but also easy to read, I would ask you to recommend it to me.

Kathleen, I like your post because I have decided to read it in a certain way, and really admire your optimism about an AI machines ability to achieve time travel before humans, let alone either achieving such a thing at all. Though I am less optimistic about this (the Tardis is not technically a machine, you know), I do think that the field of strong AI needs more people like you. Personally, regarding your actual question, I think that many machine have a far superior ability for things like mathematics, or finding the nearest public restroom, than I do, so yes, I think a lot of them have already gained an advantage over us. Do I think this will cause any trouble in the world or threaten mankind in any way? No I do not.

As for my favorite topic of study in AI, it's currently computational humor. I am writing a paper on it the same class I'm writing this blog for (Cog468), and papers in this topic are fun to read. It makes you think about humor, and its possible applications, in a different way, and you get to read about programs that do funny things, like call MIT the Mythical Institute of Theology.

Hope that answers your questions. Feel free to comment if I missed any, or you have any more or want to respond to anything I said.

No comments:

Post a Comment