# Artificial intelligence

**URL:** <https://boards.straightdope.com/t/artificial-intelligence/69349>\
**Category:** Factual Questions\
**Created:** [June 19, 2001, 3:31pm UTC](https://boards.straightdope.com/t/artificial-intelligence/69349 "2001-06-19T15:31:20Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Torkle](https://avatars.discourse-cdn.com/v4/letter/t/edb3f5/32.png) [@Torkle](https://boards.straightdope.com/u/Torkle)\
**Post date:** [June 19, 2001, 3:31pm UTC](https://boards.straightdope.com/t/artificial-intelligence/69349/1 "2001-06-19T15:31:20Z")

</div>

I’ve been messing around with AI programs for a week or so I found online and I was wondering how do ones at like [http://www.zabaware.com](http://www.zabaware.com) work? they seem sometimes to be really intelligent, but other times really stupid. thanks.

---

<div class="post-metadata">

**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\
**Post date:** [June 20, 2001, 7:09am UTC](https://boards.straightdope.com/t/artificial-intelligence/69349/2 "2001-06-20T07:09:38Z")

</div>

OK, I just played around a bit with the program there, and it is, admittedly, pretty clever. It looks like it uses a combination of methods: First, there’s some automatic responses. If you say something like “How are you doing?” (or if it thinks you said something like that), then it’ll respond with “Fine, how are you?”. Statements like this will make up a large portion of any conversation. Secondly, it uses some Eliza-style sentence restructuring. For instance, I asked it “What is the difference between a fermion and a boson?”. It replied “I know what the difference is between a fermion and a boson, but I’m not telling you.”. There’s nothing new in its response; it just re-arranged the words in my question, and stuck some filler around them.  
Finally, if it can’t figure out any other way to answer you, it’ll just change the subject. It’s probably got a whole file of phrases available like “Sounds cool. Hey, can you lend me five bucks?”.

It still has some definite limitations, though. For starters, it has no memory: Each response is based only on what you just said, not on anything you or it said before.

---

<div class="post-metadata">

**Author:** ![Pjen](https://avatars.discourse-cdn.com/v4/letter/p/3d9bf3/32.png) [@Pjen](https://boards.straightdope.com/u/Pjen)\
**Post date:** [June 20, 2001, 7:37am UTC](https://boards.straightdope.com/t/artificial-intelligence/69349/3 "2001-06-20T07:37:53Z")

</div>

Failed the Turing Test on my first question. I simply asked ‘Where are you?’ It replied ‘You’re sitting in front of your computer.’ Same answer for the question re-asked- no reformatting of answer. Seen better.
