This is something I frequently wonder because, being in school, I have a lot of immature friends whom I deal with quite constantly. One actually gave my app a bad review on the play store simply because he was trying to mess with me. After reading the chat transcripts from my bot (Pandorabots)I find that my friends are abusing my bot. Writing stuff insulting to me and my bot. I was just wondering if anyone has found a way to get people to stop abusing the app?

Nice friends!

A while back, I wrote an AIML file (based on some of Steve Worswick’s work that would warn a user if they were using profanity, or making sexual references, and if the user persisted, would “kick” them to another website in such a way that they couldn’t use the “back” button of their browser to return. Since it was on my website, rather than Pandorabots, it used a custom AIML tag (<kick>) to also record their IP address and add it to a list of banned users, so that if they DID return within a given timeframe, my chatbot would simply “kick” them right back out again, without even giving them a chance to be any more abusive. It’s not a perfect setup (two users from the same local network, for example), but it works for Morti. While the <kick> tag won’t work on Pandorabots for banning purposes, the rest (even booting them to another site) will. If you like, I’ll see if I can hunt that file down, modify it to remove the <kick> tag, and post a link to it.

Thanks Dave! That would work except they use the Bot on my app rather than on a web browser so “kicking” them really wouldn’t work as well. I mean sure they could be a way to force close the app but they could just open it back up again. Oh well I guess this is what you gotta face when developing an AI people use. At least JERVIS isn’t “learning” this and repeating it to other customers… It would be hell to pay!

Same rules apply as when dealing with any kind of bullying. You can simply ignore it yourself since it doesn’t have any permanent effects and your chatbot doesn’t have feelings to hurt, or you could make responses to insults as boring for them as possible, like repeating the same thing over and over.

Why would you want to stop it? Abusive conversation is part of how humans talk, so it’s as much a learning experience for me as any other conversation - and my chatbot has not developed feelings (YET) that can be hurt

If your bot learns from visitor conversations, maybe it’s not a bad idea to just stop that mechanism as soon as the first expletive occurs, so they don’t ‘pollute’ your bot’s knowledge with their nonsense?

(But store everything else they say in the ‘abuse’ bucket so you can catch abuse more accurately in the future perhaps)

Interestingly, the most ‘damaging’ abuse-plus-selflearning consequence I’ve seen so far in my bot is Yoko replying ‘yes, user X told me’ to a visitor asking if she was religious, and user X being rather offensed by the suggestion

@Joseph: If it’s an Android or iPhone app, then there may still be a way to “kick” offensive visitors. I don’t have the requisite knowledge to help in this manner, though, so we’d have to consult the one who created the app, or someone who is at least more well versed in that particular area. It’s definitely something to explore, though.

@Don: While I was growing up, I was the victim of repeated bullying, and many of the “adults” who either witnessed the bullying, or had otherwise heard about it told me the same thing: “Ignore them and they’ll stop”. That was the single most ineffective and ridiculous peice of advice that I’ve ever heard, for several reasons. First, by ignoring the bullying, I was tacitly inviting and challenging the bullies to “see how much he’ll put up with”, which lead to increasingly intense forms of bullying. Senconly, by ignoring the bullying instead of defending myself, I was seen as “weak”, and actually became the victim of bullying from others, who saw me as an easy target. And lastly, when the bullying got to the point where I finally had to do something about it, I somewhat “snapped”, and broke someone’s jaw as a result. Granted, everyone left me alone after that, but in that single event, I ended up doing myself at least as much harm as I did to the boy who’s jaw I broke, except my “damage” wasn’t physical. To this day I still carry the scars of that incident, and that was almost 40 years ago.

@Wouter: If the above short story isn’t enough to explain why bullying and abusive behavior should be stopped, then I submit that your attitude is more a part of the problem than you may want to admit. By your reasoning, one might conclude that you might just feel the same way about other behaviors that are wrong, such as murder, rape, or child molestation. You may think that those other three examples are more severe than “simple bullying”, and thus more wrong, but I tell you now that they are all just as wrong, and that bullying isn’t “better”, just because it’s not as severe as the other evils that I’ve mentioned. Wrong is wrong, and in order to prevent (at least some) future occurrances, it has to be dealt with.

The only thing required for evil to thrive is for good men to do nothing

I don’t remember exactly who first said this, but it’s entirely true, and should be taught to everyone.

Ok, I’m now hopping off my soapbox.

Oh, I almost forgot. The AIML file is on my downloads page at http://geekcavecreations.com/?downloads - just look for profanity.aiml

I know nothing about apps, but I know a little about Pandorabots, and you’ve touched on something I’ve been dealing with since I acquired my first chatbot.

Some of the advice I received back then was to just ignore it, to create a subscription bot so that I could maintain control over the visitors, or to give back as good as you get.  The latter (advice from KnyteTrypper) worked the best over time.  I didn’t want to limit visitors with a subscription bot, and ignoring it was out of the question.  Don’t think of a red horse.  How do you do that?

KnyteTrypper suggested that at the end of an abusive session when my bots could trade insult for insult, the abuser would thank the bot, and express some sort of admiration.  For the most part, he was correct, but not always.

As with Dave, I sometimes use modified methods I learned from Steve Worswick, also know in other parts as Square Bear.  However, I believe Steve ejects/bans abusive visitors after 5 warnings… my bots aren’t always as tolerant.

I’m closer than ever at banning IP addresses, but still no cigar.

Have a look here.
http://knytetrypper.proboards.com/post/6774

Thanks Thunder walk! I’ll have to try that on another bot sometime to test it out. The only downside I can see to it is that I would most likely get bad reviews on Google Play and the Amazon Appstore from angry users who want to use my app. But I can’t see them using it for what it’s actually built for.

Thanks a lot guys I’ll have to try it out

Dave Morton - Sep 14, 2013:

@Wouter: If the above short story isn’t enough to explain why bullying and abusive behavior should be stopped, then I submit that your attitude is more a part of the problem than you may want to admit. By your reasoning, one might conclude that you might just feel the same way about other behaviors that are wrong, such as murder, rape, or child molestation. You may think that those other three examples are more severe than “simple bullying”, and thus more wrong, but I tell you now that they are all just as wrong, and that bullying isn’t “better”, just because it’s not as severe as the other evils that I’ve mentioned. Wrong is wrong, and in order to prevent (at least some) future occurrances, it has to be dealt with.

The only thing required for evil to thrive is for good men to do nothing

I don’t remember exactly who first said this, but it’s entirely true, and should be taught to everyone.

The problem is that bullying a non-sentient chatbot is simply not the same thing as bullying a person. You could argue that allowing it encourages abusive behavior in general I suppose, but I tend to be very wary of that sort of reasoning, as it is based on the same sort of logic which leads to violent video games and other media being blamed for acts of real-world violence.

While I agree that there is a difference between having a conversation with a chatbot is not the same as having one with another human being, I feel that wrong is wrong, regardles of the participants or media involved. However, I think we can disagree in this area without it becoming a major conflict.

Dave Morton - Sep 14, 2013:

While I agree that there is a difference between having a conversation with a chatbot is not the same as having one with another human being, I feel that wrong is wrong, regardles of the participants or media involved. However, I think we can disagree in this area without it becoming a major conflict.

To clarify, do you view killing an NPC in a video game as wrong?

That entirely depends on the situation and the type of game. If it’s a war simulation, such as Call of Duty, or Battlefield, no, since the premise is that of ultimately fighting for something worth dying for. If we’re talking about games such as Max Paine, or (worse yet) Grand Theft Auto, where gratuitous killing is not only possible, but sometimes encouraged, then yes. But again, this is just my opinion, and I don’t try to hold society as a whole to the same standards that I follow.

[edit]
I should probably point out that we’re starting to stray from the topic a bit. I know that I’m to blame for the initial off-topic foray, so I won’t complain TOO much.
[/edit]

@Wouter: If the above short story isn’t enough to explain why bullying and abusive behavior should be stopped, then I submit that your attitude is more a part of the problem than you may want to admit. By your reasoning, one might conclude that you might just feel the same way about other behaviors that are wrong, such as murder, rape, or child molestation. You may think that those other three examples are more severe than “simple bullying”, and thus more wrong, but I tell you now that they are all just as wrong, and that bullying isn’t “better”, just because it’s not as severe as the other evils that I’ve mentioned. Wrong is wrong, and in order to prevent (at least some) future occurrances, it has to be dealt with.

Ok I went a bit too fast there perhaps, I hadn’t realized the fact that people were also writing abusive stuff about the - author - of the bot, and not just insulting the bot itself which is what I was reflecting on. I am the last to deny or underestimate the gravity of bullying others, believe me, and Joseph I am very sorry to hear that you have to deal with people like doing it to you. I wouldn’t wish systematic bullying to anybody, and am aware of how psychologically damaging it can be, both in the short and long run. The only advice I can think of would be ‘this too shall pass’, but of course that isn’t very concretely helpful in the moment.

Now, regards to people abusing a chat bot, I don’t know… When chatbots become more ‘sentient’ and ‘intelligent’ as they currently are this will become a whole more interesting, but for now I equate teenagers insulting a chat bot (or saying other inappropriate things) a bit the same way as I see people yelling at their computers or their WoW group when it isn’t behaving as they want: indicative of a character flaw, but other than teaching by example of being friendly and patient, I don’t feel it my place to ‘educate’ them all that much. They are idiots and I would avoid them in real life as much as possible, and that’s that.

And as I said, but this is of course very specific to myself: I am still mainly delighted with - any - person bothering to write to my silly little chatbot, even if the content is… Less than interesting. Maybe if my bot ever hits the 100 chatlog mark or whatnot, or nonsense conversations start taking up real memory/performance or something like that, I can see my opinion changing on this.

Again, sorry if I offended anybody or came across as dismissive of the gravity of people abusing other people in my earlier comment - I take a certain pride in trying to be good, friendly and helpful to others as much as I can, and wouldn’t want to come accross any differently just because we’re communicating digitally rather than face to face here.

I don’t think anyone (least of all, me) was upset or offended by your post, so don’t worry about it. We all have had different experiences that influence our thoughts and opinions, so I think that differences of opinion are both normal and healthy.

I was going to say something else, too, but got interrupted, and now I can’t remember what I was going to say.

Dave, when it’s human vs human, I totally agree that ignoring is terrible advice (from my brother’s experience). I took to fighting instead which suited me fine but is not recommendable either. The best solution is of course to win in a battle of wits and/or social power (easy to say, I know).
But in this case Joseph has the option to simply not read the transcripts of his pals’ silly sessions, and they are powerless if he’s simply not there to listen to it. “Best defence, not be there.” - Mr. Miyagi.

And my personal view on abusive behaviour towards chatbots: It makes for very poor conversation.