“Bot” is often a dirty word on Twitter, but it looks as though Twitter itself has reconciled to the concept and created its own “Welcome to Twitter” bot @TwitterSuggests. This account sends unsolicited @replies to new users, suggesting some new accounts they might like to follow. Unfortunately, it is sending some rather dirty account recommendations along with the innocent ones.

You’ll have this problem with any bot, and Twitter itself is no exception. Without human oversight, you run the risk of the bot going slightly outside of its parameters as time goes on.

@TwitterSuggests is supposed to tweet @replies to new members of Twitter. These tweets include a handful of accounts that it recommends they follow. This is obviously meant to retain new users, which is one of Twitter’s biggest challenges. However, as ReadWriteWeb first noted, the recommendations aren’t always on the up-and-up.

Although the bot is programmed to recommend follows based on an algorithm, some NSFW accounts have appeared in its @replies.

ReadWriteWeb cites one instance of the bot recommending an account that has been inactive for over 18 months as well as a pretty potty-mouthed active account as two suggestions for a new user.

And just doing some cursory exploration, we found similar recommendations going on today. For instance, @Danish786777 was told to follow an account with only 18 tweets, most of which were from August, a user who hasn’t tweeted since October and really only tweeted pleas for new follows and three polls (one NSFW), as well as a dormant account with no tweets.

These recommendations are probably going to turn people off of Twitter faster than they’ll attract new users if they’re any indication of the type of accounts the bot constantly tweets out. Dormant and/or dirty accounts won’t entice new users to embrace Twitter any time soon.

I do think the idea of the bot is pretty neat, but clearly some kinks need to be worked out. Unsolicited tweets are a dangerous thing, even if coming from Twitter itself, and the quality of the recommended accounts must clearly be improved for this to become an effective way to retain new users.