Let's Use AI to Fight Them!

Me | 3/7/2003 | BamaGirl

[BamaGirl]

Much of our activism involves sending form email or snail protest letters to particular organizations. I wonder how effective form letters (like NewsMax's PriorityGrams) are, since they are all the same -- once the recipient gets one and recognizes the format, he can throw away the rest. Personalized notes seem better, because each one has to be read individually. But the downside here is that many of us are too busy to write a note each time we are mad about something.

My solution is to create a computer program that would, given some keywords, automatically generate (infinitely many) letters that would deliver the same idea, but in a different format. We could put this on a server somewhere, and all us Freepers would have to do is push a button, and automatically get a personalized protest letter to fire off to whomever.

My question is: Do yall think this is a good idea? I would really appreciate your opinions and advice, especially from activists who have had experience in this area.

I want to use my training in AI and CS for a good cause!

[AirmanAlaska]

You set up something like that that actually works and you can bet your bottom dollar that it'll see lots and lots of use. Only problem is, it won't take long for the other side to get ahold of it to....

[El Sordo]

Or carpet bomb them with a bad movie!

[Libertarian4Bush]

ALLEN IVERSON?

[glock rocks]

so you signed up today to ask advice on how to create a denial of service server?

[BamaGirl]

Yeah, we would have to have some kind of password
protection on it.

Or better yet, have the AI be smart enough to reject
illogical (i.e. liberal) arguments outright! :D

[BamaGirl]

What? No not at all! I am thinking of creating a tool
to help get our word across, rather than ignored.
Essentially make form letters a little smarter.

Denial of service attacks are destructive rather than
constructive. It certainly wouldn't get our point
across to the people who would listen.

[PBRSTREETGANG]

So, get to work. I expect to see a prototype in 10 days.

[Technocrat]

Coding the assertions is the tricky part - once you do that, you can have the program change the order of the independent assertion trees and prune them differently for each letter. If you could come up with a way for a user to tell the program which assertions are related to each other and how, then rest is cake.

[BamaGirl]

So, get to work. I expect to see a prototype in 10 days.

Oh dear, you will have to give me more time than that! I still have to finish my thesis! :D

But I will start thinking about it.

[BamaGirl]

Coding the assertions is the tricky part - once you do that, you can have the program change the order of the independent assertion trees and prune them differently for each letter. If you could come up with a way for a user to tell the program which assertions are related to each other and how, then rest is cake.

Yeah, you are absolutely right. I think it would be cool to also provide a library of standard arguments that could be added. (You would probably need this to get enough different letters.)

[doosee]

I still have to finish my thesis!

What is the thesis subject? My advice would be to spend your time on something more constructive than camouflauging standard form letters. Spend your valuable time on a process or product that contributes to the GDP. Just my thoughts.

[BamaGirl]

What is the thesis subject? My advice would be to spend your time on something more constructive than camouflauging standard form letters. Spend your valuable time on a process or product that contributes to the GDP. Just my thoughts

Thanks for the advice. I think you are right. I just got so excited about FR and having people like yall to talk to. It still sounds like a fun problem to work on in my spare time...

My thesis is all about finding out what is formally required (in terms of languages, inference mechanisms, etc.) to make a database/knowledge base easy to change. So say Delta Airlines wants to change their schedule of flights -- how should that database be written so that it is easy to update? The idea comes from human dialogue, where people easily make updates to others (e.g. "forget I said that", "I meant X not Y," etc.) to change what they said, with minimal effort.

[doosee]

So say Delta Airlines wants to change their schedule of flights -- how should that database be written so that it is easy to update? The idea comes from human dialogue, where people easily make updates to others (e.g. "forget I said that", "I meant X not Y," etc.) to change what they said, with minimal effort.

You should read ( or may have already) about neural network systems, or systems that learn. It has been around since at least the 70's and is used in game technology, military simulations, and general industry. In my very simple terms, if you have a massive database stored with logic and logical expressions, then when one would type or say something that defies logic, the neural system would help to counteract that position. I am also a student of AI and was one of the early members of AAAI. The org. may have changed the name now, not sure. Good luck on your topic.

[BamaGirl]

Hey doosee,

That's so awesome that you were(are?) part of AAAI! I am too!

Yeah NNs are neat, but they aren't "transparent" in that you can look at them and see what they are doing. Instead I am using logic and heavy doses of nonmonotonic reasoning to get my results. It is really cool. Once you get the hang of logic, it is amazing what things you can do with it!

BamaGirl