Books for an artificial intelligence paper

[LastLfan]

Hey guys, I'm writing my senior thesis on the ethics of artificial intelligence and I need some good thought provoking books and potentially movies. I prefer fiction cause its my belief one man's entertainment is another's doctrine. It has to have some sort of reasoning behind it not just pure action(so more i, robot than terminator). Thanks guys

[nillapoet]

I.Robot. was good.

[Ante Bellum]

It's a bit of an odd suggestion, but the game Infinite Ocean (just an online point-and-click, nothing intense, and there's a walkthrough) might work.

[ClosetOtaku]

For what it is worth, I've dabbled in the AI field over the years. Fiction will only get you so far with ethics, IMO. I think this for two reasons:

a. Fiction rarely recognizes constraints; that is, you might end up addressing a problem that is not likely to emerge, maybe ever;
b. So much of what is billed as AI today is marketing rather than true AI, and if you want to honestly address problems emerging from the use of AI, you want to find some real AI issues

So, assuming you are SERIOUS about this, and have all sorts of time on your hands, consider the following --

Two books immediately come to mind: "Computer Power and Human Reason" by Joseph Weizenbaum, and "The Age of Spiritual Machines" by Ray Kurzweil. They are both well-respected non-fiction books that have stood the test of time (Weizenbaum's was written in 1976). They address concerns that, in my mind, are still valid -- mostly because we haven't made all that much progress in true AI in the past 30 years (opinions vary on this, but I think most academics agree we have not moved as quickly as we thought we would).

But, this being so late in the semester, and time is of the essence. So maybe fiction it is. Have you considered "Blade Runner" or its source "Do Androids Dream of Electric Sheep"; "The Moon is a Harsh Mistress"; the computer HAL in "2001" (although we only learn why HAL became psychopathic in the sequel "2010"); and perhaps the recent "Her"? Lots of books out there, these are just off the top of my head, if I think of more I'll post them here.

I'll go out on a limb and say even "Frankenstein" (the book, not the movies) might give you some interesting food for thought.

The issue I have with "I, Robot" is that Asimov constructed his Three Laws, then asked the question, "How would the Three Laws play out in real life, outside of the sterile laboratory in which they were conceived?" It is not so much a series of short stories about AI as it is an exploration of lexicon and logic.

Some terms worth Googling: Turing Test; Complex Adaptive Systems; Emergent Behavior; Neural Networks.

All of this is meant to give you some ideas about how AI works -- you can likely develop ideas about the ethics of it from there. I think that might be more fruitful than trying to expound on a fictional strawman, but Your Mileage May Vary.

Good luck!

[LastLfan]

Actually this is a paper I have all next year to work on, I just want a head start. Thanks for the advice, I want to list a couple of fiction sources because some of them do address real issues but yes I will definitely turn to non-fiction as well. I'm also gonna have to define ethics in order to set up an arguement of any sort. I'm definitely seious about this and have already picked up do androids dream of electric sheep. I want to use fiction to explore the differing possibilties and how different people have ideas on how ai will affect us. So its going to be mostly speculative

[Mr. SmartyPants]

I guess you can look up like Thomas Hobbes or something. Say something about how we're artificial intelligence as well or something.

[ClosetOtaku]

It is great you have some time to work on this, and better that you are getting a head start! That's very admirable, and I hope it pays dividends in the quality of your project.

I've done some work on the intersection of AI and medicine (my field of study is Medical Informatics). While some of the questions raised by fiction (and science fiction) on the topic of AI are interesting, many questions we are currently confronted with in medicine may have some relevance to your work. For example:

- Some people relate better to computer interfaces than real people, particularly in the medical realm. Would you trust a computer's expert opinion over a human Doctor's?
- Would you want to know how the medical AI was programmed? What if it were programmed by a Big Pharma or an Insurance Company, would that change your confidence in its accuracy or neutrality? Should disclosure of such information be mandatory?
- Would you consent to having an AI monitor your compliance with diet, exercise, medication, etc. if, in exchange, you received a discount on your medical costs? (Warning: in the future, this may not be optional.)

Now, in reality, a lot of AI in medicine has little to do with patient interaction -- much of it pertains to pattern recognition, interpretation and classification of large amounts of textual historical data, advanced work in human genomics, and so on. There aren't too many ethical issues directly associated with these per se, although peripheral issues are present: I was on a teleconference recently with a number of physicians who were convinced that, within a decade, your entire genome will be part of your medical record. There is potential room for abuse (watch "GATTACA" sometime), but this has less to do with AI and more to do with the improvements in technology that make such situations possible.

Your approach sounds very interesting, and I think you have a great and fascinating journey ahead of you!

[LastLfan]

Thanks for the medical insight, I had never even thought of that. Considering this is only 20 pages I have to consider strongly what to include and that is definitely something to address. I think one point of ethics i want to get at is when making ai becomes less about helping humanity and more about playing god and trying to create life

[ClosetOtaku]

Speaking of the Turing Test... in the headlines yesterday: http://io9.com/a-chatbot-has-passed-the-turing-test-for-the-first-ti-1587834715

[ClosetOtaku]

Aaaannnd... Counterpoint. https://www.techdirt.com/articles/20140 ... tter.shtml

This goes to show a few things:

1. The Turing Test (despite what this latest article says) in a big deal in the world of AI milestones.
2. There is controversy over exactly what constitutes a successful "pass" of the Turing Test.
3. If the test had been passed in, say, the 80's, maybe we could claim to be farther along in AI technology than we are now.
4. As I mentioned in an earlier post, most of what passes for AI today is marketing, not true AI.

Food for thought....

[blkmage]

One of my favourite recent treatments of the philosophical implications behind AI from a theoretical computer science point of view (specifically computational complexity theory) is from Scott Aaronson's survey paper on the perspective of computational complexity in philosophy (as opposed to computability theory). There's a section with a slightly more detailed treatment of the Turing test explaining what it means, what it's intended to do, and the implications on what it means for a computer to be able to "think".