Stephen Smith's Blog

Musings on Machine Learning…

Archive for February 2019

Social Media Bots

leave a comment »

Introduction

In my novel “Influence”, the lead character J@ck Tr@de spends a lot of time creating and improving Social Media Bots. I thought in this article I’d spend a bit of time providing some background on these. Social Media Bots weren’t made up by me, they’ve been around for a while. It is estimated that 15-20% of all social media accounts are really Bots and that 15-20% of all posts on social media sites like Twitter and Facebook were created by these Bots.

For anyone interested, my book is available either as a paperback or as a Kindle download on Amazon.com:

Paperback – https://www.amazon.com/dp/1730927661
Kindle – https://www.amazon.com/dp/B07L477CF6

What is a Bot?

Bot is short for Robot and really means a computer program that is pretending to be a real human being. Early bots were easy to identify and rather simple, but over time they’ve become more and more sophisticated, even to the extent of being credited with getting Donald Trump elected as president of the USA.

The original Bots were spambots, which were programs that just send out spam emails. Basically hackers would take over people’s computers and install a program on them (the spambot) which would then send out spam to all the contacts on the poor victim’s email. Programmers found these quite effective and took the same idea to social media.

Most of these Bots are quite simple and just work to advocate some idea by posting from a collection of human created messages. They can be trying to influence political views, direct people to dubious websites or perhaps just make people mad for the fun of it.

There is an interesting website, Botometer, that will analyse a Twitter account and score it to see if it’s a bot. I ran it on all my Twitter followers and quite a few got a score indicating they were Bots.

Bots Get More Sophisticated

Like any computer programs, Bots keep coming out with new versions getting more and more sophisticated. They now create quite realistic Internet personas with photos and a history. If you look at such a Bot’s Facebook page, you might be hard pressed to tell that it doesn’t belong to a real person. Creating social media accounts is pretty easy with very little verification. You just need a valid email account and need the ability to respond to the email that is sent to ensure it is you, plus perhaps fool a simple captcha tool.

Another newer Bot is the so called ChatBot. These are programs that can carry on a conversation. They can use modern sophisticated machine learning algorithms to carry on a conversation on a topic like providing movie reviews. Many companies are trying to deploy ChatBots to automate their customer service. Companies can purchase ChatBot kits that they can customize for their own customer service needs. Often companies use ChatBots to handle their social media accounts. A major large company can’t answer all the Tweets and Facebook posts it receives, so they automate this with a Chatbot. Sometime this is effective, sometimes it just pisses people off. The feeling is that people getting some sort of answer is better than no answer.

The developers that create Social Media Bots took this same technology and incorporated it into their Bots. Now these Bots don’t just post canned messages, but can also carry on limited conversations on these topics. Often political campaigns employ these to give the impression they have far more support than they really do. If you post a comment to a news article on Facebook, often you get responses almost right away. Often most of these responses are actually from Social Media Bots using ChatBot technology. The Russians really spearheaded this in the American 2016 election campaign.

As Machine Learning and AI technology gets more and more powerful, these Social Media Bots get harder and harder to distinguish from real people. Especially given the low quality of posts from actual real people. When a corporation uses a Chatbot for technical support, it will identify itself as such and often has an option to switch to a real person (perhaps with quite a long wait time), but when you are on Social Media, how do you really know who you are talking to?

In my book, Influence, the main character, J@ck programs his Bots to both network and to modify their own code. As it is, Bots behave as viruses and spread maliciously from computer to computer. The current Bots tend to rely on volume to do their damage. But as in Influence, perhaps the Bots will start to coordinate their actions and work together to accomplish their goals. Given the number of computing devices connected to the Internet, a successfully spread Bot could harness tremendous computing power to spread its Influence. Applying new algorithms for reinforcement and adaptive learning, the programs can get more and more effective out in the wild without requiring additional coding from their creators. Is it really that far fetched that this network of Bots couldn’t become aware or intelligent in some sort of sense?

Summary

Twenty percent of users and twenty percent of posts on Social Media are via automated Bots and not created by real people. Should you believe what you see on Facebook, should you be influenced by all the tweets you see going by on Twitter? Are your thought processes critical enough to filter out all the automated noise that is being targeted at you? Are your consumer decisions on what you buy or your political decisions on how you vote being controlled by all these Bots? This is definitely something people should be aware of you should be aware of this and don’t just believe it all.

Advertisements

Written by smist08

February 5, 2019 at 9:38 pm