Hi everybody

I recently found this website, and I’m glad to find so many enthusiastic people here. But I’m curious about the chatbots you all make. What is the purpose of those bots? Are they meant to perform a certain task (selling train tickets for example), are they question-answer systems, or anything else?

Also, what is (in 1 or 2 lines) your general approach? Do you all use AIML? And if not, what then?

~Mark tM

Hi, Mark, and welcome to chatbots.org.

While I can’t speak for everyone else (or anyone else, for that matter), I can tell you about my bot, and what it’s for. First off, my bot, Morti, is one of only a very few AIML chatbots here. His purpose is to just carry on an entertaining, whimsical, and somewhat light-heartedly sarcastic conversation with those who visit him. Morti’s also a sort of “free-form training vehicle” that I’ve been using to learn certain techniques in the areas of PHP/web programming, AIML construction, and even, to a certain extent, other avenues of research into AI. The community here is pretty eclectic, when it comes to AI research, and it often seems that there are more projects being pursued than there are members to pursue them. Feel free to poke around in the threads here (especially the “Projects” forum), and if you see something that you wish to comment on, don’t hesitate to do so.

Welcome Mark,
For my own part I have 2 components related to Chatbots.

- JAIL (JavaScript Artificial Intelligence Language) a framework for developing Chatbots and AI projects.

- Skynet-AI a chatbot built with JAIL. I use it to explore AI concepts, compete in chatbot question answering competitions, and hold on-line conversations with a wide variety of visitors.

What is the purpose of those bots?
Chatbots can often be thought of as any other hobby, like stamp collecting or gardening. In my case I use it to test AI concepts and to see if there may be a commercial benefit to the systems I am creating.

Also, what is (in 1 or 2 lines) your general approach?
Skynet-AI is created with JAIL. It is designed to be small and fast, running with limited resources on a wide variety of platforms. The AI runs in the client and makes extensive use of web resources.

Hello there, Mark tM,

Merlin above has given a good answer-format, so I’ll borrow it.

What is the purpose of those bots?

Like Merlin, I use the http://www.scn.org/~mentifex/AiMind.html chatbot to “test AI concepts” and to implement an http://mind.sourceforge.net/theory5.html (Theory of Mind) for AI that I labored over for many years. Other people could adapt the Mentifex AI Mind for such purposes as answering questions from a specific knowledge base (KB), but I keep busy just implementing the basic Strong AI engine. The ultimate goal of these AI Mind bots is to install them as the minds and personae of autonomous mobile robots.

Also, what is (in 1 or 2 lines) your general approach?

Gee, Merlin must be my guide and role-giver, because I also use JavaScript, but only for the tutorial version of the True AI chatbot. Just this morning, as usual, I was coding the AI Mind in the traditional robot-programming Forth language. As you may see from the AiMind thread at http://www.chatbots.org/ai_zone/viewthread/240/ my “independent scholar” approach to AI has generated a bit of controversy over the eighteen years that I have been coding the True AI since 1993, but I just keep plugging along and releasing my open-source AI code into the public domain.

The purpose of my work (subbot.org) is to test multi-agent theory applied to chatbots.

The general approach is to send input (asynchronously, in parallel) to agents (currently around 25, including ALICE, MegaHAL, Link grammar, Montylingua, WordNet, Wolfram Alpha, Wikipedia, Google Calculator, my own logicagent...), then select the highest scoring response. Each agent scores its own response; the scoring (and responses) can be changed (at runtime) through feedback from the user.

Wow. Must..resist..urge..to..start..detailed..discussion..with..everybody.

@Dave: I know what you mean, hobby projects are awful for your free time.

@Merlin: JAIL sounds interesting, do you have more information about it anywhere?

@Arthur: Strong AI? Cool, although it sounds optimistic By the way, are you referring to the Theory of mind in which the agent tries to deduce the beliefs, desires and intentions of the user?

@Robert: Clever, using the work of others Although, it is not really a multi-agent system in the sense that the different software-agents (not to be confused with virtual agents)  do not interact with each other.

Also, I just noticed that I haven’t even properly introduced myself I’ll do that now in the correct section.

~Mark tM

Skynet-AI
http://www.chatbots.org/ai_zone/viewthread/316/

JAIL
http://www.chatbots.org/ai_zone/viewthread/317/

Mark tM - Sep 14, 2011:

[...]@Arthur: Strong AI? Cool, although it sounds optimistic :) By the way, are you referring to the Theory of mind in which the agent tries to deduce the beliefs, desires and intentions of the user?

No, not the so-called “child’s theory of mind”.

http://code.google.com/p/mindforth/wiki/BrainTheory

is my attempt to explain how the brain works and

http://www.scn.org/~mentifex/AiMind.html implements the theory.

Hi Mark, welcome to the forum (I read your intro post befire this one )

ALF is a project I’m working on, there is a thread in the Projects section that tracks my progress, thoughts and other things with regard to the R&D.

What is the purpose of those bots?

I’m not sure if ALF really classifies purely as a chat bot, as while that’s a “goal”, it’s not all that I plan to do with it.  Someone in another post on here coin the phrase “thot bot” (thought bot) and I would suggest that is more what I am aiming to achieve.  Really, I’m just very interested in the whole field of AI, and have been messing with “bots” and robotics since I was 8-9 (I made my first chat bot on a BBC Micro hehehe).

Ultimatley, I would like ALF to be able to “think” and solve problems, perform deep QA, and be able to hold a conversation with “personality”.  However, thats a mammoth task and I’m really quite happy just messing around trying new things, learning new tricks and seeing what I can do

Also, what is (in 1 or 2 lines) your general approach?

I tend to go with the flow to be honest, having a clear and defined plan of action past the starting post in the field, is in my opinion difficult and naive, as you are planning completely against the unknown.

I’m well disciplined in a number of technological fields, telecoms, mathmatics, physics, game AI (I used to write games you know ) etc etc which kinda helps

Basically, if I’m missing it, and I need it, I build it with regard to ALF

Mark tM - Sep 14, 2011:

@Robert: Clever, using the work of others :P Although, it is not really a multi-agent system in the sense that the different software-agents (not to be confused with virtual agents)  do not interact with each other.

Sure they do. The In Soviet Russia agent, for example, sends strings to the Link Grammar agent, same as a user might. “What is the object in ‘You can always find a party’?”, etc. When the ISR agent gets the object it makes it into the subject, tries to change the verb agreement if necessary, and puts ‘YOU!’ after the verb…

Hi Mark.
Let me jump in:
What’s the purpose of my designer: provide na easy entry point for people to build their own personal/pc assistent, sales bot,....
My approach: resonating neural networks + good UI’s that make it easier to use.

The purpose of a chatter robot is to practice a craft.

Why are you practicing a craft? I bet my answer is different from yours :)