Bot Contest

I think what they look for is the appearance of understanding. It shouldn't matter whether it's specifically human understanding or not, though some judges might weight decisions in that direction

I've been looking at the transcripts from last year, and I think there are a few bots here that are much better than Alice. At least four of the finalists should come from the Forge.
What do you guys think that causes most impression that the bot really understands what it's saying?

If NotKirby makes it to the finals, I'll buy a round at Dogh'd's.

If NotKirby makes it to the finals, I'll drink that round.

Dont forget a round for me!

So it's not like a Turing Test where the judges don't know at first whether or not they are talking to a bot?

The idea of the Loebner is classical Turing. There are a couple of humans with the finalists. But they aren't biased against bots that state clearly they are bots (Alice is one of those). Pressumably, they won't have anything either against bots that believe (rightly or otherwise) they are gods.

Sounds great! I often wondered what kind of connection they thought there was between being intelligent and being human.

Personally, I think any sufficiently advanced bot here could pass the test if there's a way to recognize plug-ins. Maybe there already is and I don't know it.

We would need to create more plugins, of course, and more careful lists.

There's getting to be enough public plugins, maybe they should be separated by letter like keyphrases are.

It's true that we need more plugins. I plan to write a few public ones when I have the time for it. My goal is to write enough plugins to make it possible to include random sentences in a bot that make sense most of the time, where all the elements are plugins.

Random sentences are what I do best. I've written a lot of programs to make random sentences of various types, as well as random D&D type items, critters, etc. In fact they're about the ONLY thing I can program. This is why most of my bot's stuff is random, i.e. she responds to the "I have a" rather than the item the person has. Only actual subjects she has are things everyone talks about, i.e. weather, family, job. But she doesn't have any set responses to those, for instance every time she tells you her job it'll be a different one. Not very consistent I know, but it makes for funny as hell transcripts, which was the whole point in making her to begin with.

What I meant was having the bot recognize a word or phrase as belonging to a certain plugin, and responding accordingly. I don't think we have the capability for that. Yet.

BTW, thanks much to those of you who have been thoroughly humiliating my bot today. She obviously is in need of human interaction with someone besides me. I really do appreciate your time. Thanks.