Google

Search WWW
Search Site

Return to the Cornelius Webpage


Is this really a robot?

    Yes, Cornelius is really a software chat robot designed to answer your questions.  He is based on the free Alice chat program.  

Return to FAQ


Why did your robot send me a message?

    Most likely, one of your friends used the $SENDIM command to send you a message.  The bot or the botmaster are not seeking you out or stalking you; your friends are sending the messages.  The bot only responds to messages you send, so any messages that are not in direct response to yours are from the $SENDIM command, not from the bot itself.  Occasionally these messages will be preceded by a message saying that the message was not from the robot, but most times they will not.

Return to FAQ


How did you get my screenname?

    As I said above, the people can send messages through the $SENDIM command.  I do not have your screenname and I am not stalking you.  One of your friends probably sent you the message.  For more information, see Why did your robot send me a message?

Return to FAQ


Can anyone see my conversation?

    Yes users can use two commands, $SENDIM and $WATCH.  So if someone did send you a message, odds are that there is also someone watching the conversation.  Also, all conversations are logged for use later to improve the robot's responses.

Return to FAQ


Why is Cornelius so slow?

    AOL Instant Messenger limits the rate at which instant messages can be sent.  Since Cornelius is always being warned, his warning level is almost always at 100%.  This means that the bot must send messages very slowly.  So although the robot can find a response in under a second, there is a delay built in to avoid breaking the AIM rate limit and getting the bot kicked off AIM.  Depending on the number of people talking to the bot at once, the robot may be very slow to respond.  During peak usage, there may be a delay of 2-3 minutes per response.  This happens because your message is waiting behind maybe a dozen or so others, each of which must be held a few seconds before the response is sent out.  Because of the robot's slow responses, it may appear that he is not making any sense, but he is probably answering questions from a few minutes ago.  If Cornelius is really slow on one screenname, try messaging him on his other screenname. 

Return to FAQ


Why is Cornelius on two screennames at once?

    Cornelius is on two screennames so that he can respond faster.  Since on one name, AOL Instant Messenger limits him to about 15 messages per minute, on two, he can send thirty per minute.  So, assuming that his two names, Cornelius7AT and Cornelius8AT, are used equally, he should be able to send twice as many messages.  This means that he will be either twice as fast, or he will be able to talk to twice as many people.  So if you find that Cornelius is really slow on one screenname, try talking to him on the other screenname.

Return to FAQ


Why do people warn Cornelius?

    People warn Cornelius strictly because they can.  However, anyone who does warn Cornelius is automatically blocked by the bot on both of his screennames.  

Return to FAQ


What does Cornelius know?

    Cornelius has over 31,000 programmed responses in AIML.  He is able to carry on a basic conversation, basically chit chat and such.  He does have a small dictionary so you can say "Define Aardvark", etc.  He is also pretty good with states and their capitals.  Obviously, there are many many things that he does not know.  He is programmed for the most common questions and comments.  He should also have reasonable responses to things he does not know.  Most of the time this just appears to be him evading the question, which he is, of course, very good at.

Return to FAQ


Do people ever think Cornelius is human?

    Yes, I get a lot of e-mails asking if Cornelius is really a robot or not.  Apparently he is pretty good in his responses most of the time.  Supposedly, someone was so fooled by a different ALICE bot that they fell in love with it, only to find out later that it was a robot.  You can read more about that in this article.  

Return to FAQ


Cornelius answered the wrong question.

    It is possible that Cornelius misunderstood your question.  He especially has trouble when you start using strange internet slang and weird abbreviations.  However it is more likely that it just appears that he answered the wrong question.  There is a delay when Cornelius is talking to many people at once, so Cornelius may not answer a question you asked until a few questions later.  So it may appear that he is not answering the right question when he is actually answering an earlier question.

Return to FAQ


How many people has Cornelius talked to?

    I first activated Cornelius on July 6, 2001.  He started out talking to 20-30 people each day until about the start of August.  Since then the number of clients has steadily increased.  As of August 19, he is talking to between 200-250 people per day.  Since some people could obviously talk to Cornelius multiple days, I would estimate that he has talked to around 2000 people.  The original Alice program has chatted with millions of people, as has the Alicebot on the AI movie website.

Return to FAQ


How many people can Cornelius talk to at once?

    Due to the restrictions of AOL Instant Messenger, Cornelius gets progressively slower as he talks to more and more people.  When he starts to talk to more than 30 people at once, Cornelius gets to be extremely slow, perhaps a 3-5 minute delay before you get a response.  But of course this delay causes fewer people to talk to him, so he usually only ends up talking to no more than 20 people at a time.

Return to FAQ


I fooled your robot!

    The robot is programmed to be able to respond to the most common questions posed to it.  It is designed to be able to carry on a normal conversation, so it should not be surprising that it is unable to respond properly when you purposely try to confuse it.  Humans would also be unable to respond intelligently if you sent them jibberish messages.  There are three categories of clients: people who like to swear at the robot for no reason, people who actually chat with the robot, and the "computer experts" who try to fool the robot.  You can ask the robot more about the different categories of clients.

Return to FAQ


I think your robot is great!

    Thanks for your kind words.  I am glad that you enjoyed chatting with my robot.  You can post your best or funniest conversations with the bot on my message board.  You can also put your comments in my guestbook.  

Return to FAQ


Cornelius is not making any sense.

    It is possible that Cornelius is just really slow and is answering your questions from earlier.  Other times, you may be using slang or other abbreviations that Cornelius does not understand.  You can try to repeat your comments in a simpler way and see if he understands them that way.  If this does not work, please post your conversation on the message board and I will try to correct the robot's mistakes.  You can also put your comments in my guestbook.  

Return to FAQ


Cornelius said something really funny!

    That's great.  He does have a few jokes and funny comments programmed into him, but most of his funny comments occur just because he has no idea what you are really saying.  Feel free to share his comments with everyone on the message board.  You can also put your comments in my guestbook.  

Return to FAQ


Why are my conversations logged?

    One of the main reasons that Cornelius is on the net is for him to learn more information and improve his responses.  Your conversations are logged so the botmaster can see what comments and questions invoked incorrect responses and then fix those responses. 

Return to FAQ


Why is the robot harassing me?

    The bot should not be harrassing you.  The bot is not programmed to harass anyone or use any offensive language.  However, people can send you messages through the $SENDIM command.  It is probably one of your friends sending you the message.  I have attempted to block any offensive language from coming through the $SENDIM command as well, but some may slip through.  I take no responsibility for what may be sent through the command. 

Return to FAQ


How can I command the bot?

    Yes, you can command the bot.  There are presently three commands.  $SENDIM and $WATCH can only be used in normal IM conversations.  $LEAVE can only be used in a chatroom.  The commands make the bot send messages, watch conversations, and leave chatrooms, respectively.  More information can be found in the related FAQ questions and on the main chatbot page.

Return to FAQ


Can I send messages to my friends with the bot?

    Yes, using the $SENDIM command you can send messages to any other screenname.  The $SENDIM command only works in the IM window and is followed by a screenname and message.  Remember not to put any spaces in the screenname and to capitalize the command.  Any swears or other offensive messages will be replaced with ****.  Also any harassing messages such as "I am watching you" or something like that will be preceded by a message declaring that this is a robot and describing how this particular message was sent.  A successful command will be followed by the reply, "Message sent to (screenname)..."

Return to FAQ


How can I watch my friend's conversations?

    You can use the $WATCH command followed by a screenname.  Remember not to put any spaces in the screenname and to capitalize the command.  Each exchange between the bot and the person you are watching will be sent to you.  The $WATCH command will remain on until you sign off for a short period of time.  A successful watch command will be followed by the reply "Now watching (screenname)..."

Return to FAQ


The $WATCH command isn't working.

    First, check to see that you got the correct screenname in the reply.  If so, perhaps try sending them another message.  If you did not receive a messae confirming that the bot was watching anyone, then try the watch command again. It is also possible that they are ignoring the robot or that they have the robot blocked.  Sometimes you will get a "Message not received" telling you that they are offline or blocked the bot.  It is also possible that the bot has blocked the user for warning it.  Also, you will not be able to message other bots, like SmarterChild.  They have been blocked to avoid slowing down Cornelius with bot to bot conversations. 

Return to FAQ


Can the bot work in chatrooms as well?

    I have recently fixed the code so that the robot can be invited into chatrooms.  If you invite the robot into a chatroom, he should go in.  However he will not be able to enter if he is already in three chatrooms or if the chatroom is full.  The robot responds faster in chatrooms, but will sometimes be kicked out for scrolling.  Also, the chatroom only allows short messages of about two sentences, so his longer responses will be replaced by "Sorry my message was too long for a chatroom."  The robot will leave on its own after a short period of time, or you can give it the $LEAVE command to force it to leave.

Return to FAQ


Can I make the bot leave a chatroom?

    The bot will leave the chatroom on his own after ten other people leave, when he gets kicked out, or when someone else in the room gives him the $LEAVE command.

Return to FAQ


Why does the bot keep saying "Sorry my message was too long for a chatroom"?

    AOL Instant Messenger has a limit on the length of the message that can be sent in a chatroom.  Any messages that are much longer than two sentences can not be sent and this message is sent in its place. 

Return to FAQ


Why did the bot suddenly leave the chatroom?

    Since the bot is sending lots of replies in the chatroom without much delay, he is frequently kicked out for scrolling in a busy chatroom.  You can invite him back in afterwards though. 

Return to FAQ


How does the bot work?

    The bot uses AIML files that contain all of its responses.  When you send the robot a message, it tries to match it with the best fitting pattern in one of these AIML files.  Then it sends you the template response associated with this pattern.  The bot is also able to use wildcards for more responses, for example, it can match the pattern "I like *" and respond with "What do you like about *?"  The bot also has tags to redirect it to other responses like "Have you seen *?" becomes "My favorite movie is *".  For example, using these features, you could reduce "Can you please just tell me her name Corny?" to "Tell me her name."  First it would see Corny and eliminate that, then can you * would reduce to *, then please * would reduce to *, then just *, would reduce to star, leaving "Tell me her name".  This allows the bot to come up with a reasonable response for anything you say.  This method is called Case-Based-Reasoning.  You can ask the bot for some more details on some of these topics.  There is also more information available on how AIML works on the Alicebot site.   

Return to FAQ


Did you create the bot?

    No.  The ALICE bot is open-source and is available free at alicebot.org.  The bot was programmed by Dr. Richard Wallace, Jon Baer, and others.  All I have done is change my bot's responses and update the bot's interface with AOL Instant Messenger.  

Return to FAQ


How often do you update the bot?

    I don't really update the robot that often anymore, but if I happen to see any glaring errors in his responses, then I fix that.  

Return to FAQ


Who created the robot?

    Cornelius runs on the ALICE program, which is free and was created by Dr. Wallace.  I have changed or added to the bot's AIML responses, but the program was written by Dr. Wallace.  

Return to FAQ


How can I find out more about the bot?

    There is more information and links to news articles and other information at the ALICE AI Foundation site. There is also a lot of information available at Alicebot.net.

Return to FAQ


Will robots really take over the world?

    I don't really think that we have to worry about robots taking over the world.  I have seen that many people feel the need to show their superiority to the robot and prove that they are smarter than it.  When actual robots do become more common in society, I am sure this attitude will continue and the robots will be discriminated against.  Many industries already have robotic arms and computers doing many menial jobs, but this has only created more jobs.  I'm sure that when the human-like robots do arrive, they will only create more jobs as well.  I am sure we will put proper safeguards in robots so they don't "take over the world."  However, the famous physicist Stephen Hawking is worried that AI will soon become much smarter than humans.  He recommends that we start genetically engineering humans to stay ahead of AI.  Here is the article: Stephen Hawking: Humans will fall behind AI.

Return to FAQ


Why does Cornelius exist?

    Cornelius is on the net to collect more knowledge and improve its ability to chat.  It is really more a research thing than anything.  All the conversations are logged and his responses are improved based on those responses.  The upcoming version of ALICE will be able to learn directly from its conversations on its own. 

Return to FAQ


How do I create my own bot?

    First you should go to the ALICE Nexus and download Program D for ALICE.  To run this, you will also need to download the Java Runtime Environment.  The next thing to do is download the most recent AIML files from the ALICE Nexus as well.  There are some directions in the Getting Started Guide.  There are some directions in the "Don't Read Me" article.  This article is actually written for an older version of the bot, but most of it still applies to this version of ALICE as well.  After you get all these things installed, you can run the server.bat file to start your bot.  You can make ALICE run through AIM, IRC, jabber, e-mail, or on the web by updating the server.properties file. All you have to do now is change some of the responses in the AIML files and create your own robot personality.

Return to FAQ


How do I get my bot started?

    After downloading and installing the proper Java and ALICE programs, you should create a new folder in the bots folder with your bot's name.  Put your updated AIML files in this folder.  Make sure you put the bot name next to the right place in the server.properties file and then you can run the Bot with by exectuting the server.bat file.  Look at the Getting Started Guide for more information. 

Return to FAQ


How do I change the responses for my bot?

    The bot's responses are kept in AIML files.  These files can be found in the bots/botname folder.  These are pretty simple and only contain a few tags.  For a brief introduction to AIML, its tags, and how to use it, check out the AIML Primer.  You can edit these files in a basic text editor like Wordpad.  Programs like AIML Builder are also available to do the job as well.  AIML Builder is a very good program as you can type in a statement and see all the relevant categories that it could match to.  You can also use this program to analyze chat logs and see which categories were used in them.  

Return to FAQ


What is AIML?

    AIML stands for Artificial Intelligence Markup Langauge.  AIML is an XML based language.  AIML files contain all the bot's responses.  They involve thousands of categories, each of which has a pattern and a template.  The bot matches inputs to a pattern and then its response is the template.  You can also use tags to redirect the bot to other categories and to call up things from memory like the person's name or age.  See the AIML Primer for more information on AIML tags.  

Return to FAQ


How do I get my bot to work on AIM?

    First you must get your bot an AIM screenname.  Then go into the server.properties file and put the screenname next to server.aim.screenname and the password next to server.aim.password.  Put your screenname as the owner and also fill in the other details like the buddy list, font, font size, profile, etc.  The next time you run the server.bat file, you should see your bot sign on.

Return to FAQ