When I started The Environment Site we had a feature on our discussion forums called eCoBoT.  This was a chat plugin.  It used fairly basic AI to enable you to have a conversation with it.  You would post a message, eCoBoT would make a reply based upon this input, and so on.  It was unlikely to ever pass the Turing Test (although some members did think it was one of the admin team pretending), but the results were often very funny.

That was several years ago and this kind of thing hasn't really moved any further away from the fringes.  I was reading today however about a Twitter Bot.  This isn't a TwitterBot in the sense of having something merely retweet or follow particular accounts.  This is a social bot that can actually hold conversations.  Apparently anyway.

The research had it's origins in a scepticism around claims by social media 'experts' that they could grow clients' Twitter networks and increase engagement with their followers.

"A lot of people you can hire now say they are really good at community engagement," says Tim Hwang, one of the authors of a research paper describing the socialbot experiments. Hwang and his colleagues wondered, "Can we measure those claims?"

They joined the Web Ecology Project, in particular a competition to see which team of researchers could gain the most @replies on Twitter.  There was nothing in the rules to say the process could not be automated, so they set about trying to win the competition by using bots.

To do this they followed 2,700 Twitter users over the course of 54 days.  The first 33 days were the control period, so no social bots were deployed.  The following 21 day period however saw 9 bots cut loose, one for each target group of 300 users.  Each bot was instructed to do things like retweet messages and introduce users to one another.


Interestingly it seems that many didn't seem to notice they were interacting with a bot.  The experiment found that each bot account attracted on average 62 new followers and received 33 incoming tweets, split between @mentions and retweets.

What's even more interesting is the impact of the bots on the target group.  Compared to the control period, the bots involvement saw a 43% increase in follows.  One group saw a whopping 355% increase.

With crowdturfing on the increase this is indeed an interesting development.  Can you see any positive use of this technology?  I'm not sure I can but look forward to hearing your comments.