Been in a fruitless loop of a conversation about your Cab complaint or a flight refund on a social media platform? Chances are you have been speaking to an AI powered bot employed by the respective cab service provider and airlines to help with some tasks which have the same or similar questions as well as solutions. Worst still, you may be interacting with a fake account. Advanced ML is making the bots seem more and more humanlike by adding nuances which correspond to human behaviour in a very convincing manner and hence these bots are hard to identify. We take a look at ways in which one can identify bots.
Social Media Bots
Take a good look at the user profile. You may notice the lack of a profile picture and nothing in the bio either. If the bot has a profile picture it is likely to be stolen from the web. The account name could be a randomly generated name too. Social Media bots, especially those on Twitter, are assigned a particular task and you may find them to go on and on about a particular topic (task assigned). This includes posting the same links or content. Moreover, the language they use sometimes gives out their identity. Even though ML has made a lot of progress it is still challenging for bots to speak/write in the human language. Social Media bots are also over active. Non-stop 24×7 posts from an account could indicate they are a bot too. Such accounts are usually also followed by a lot of bots.
Asking the chatbot an emotional question will generate a standard reply or a standard reply in two or more ways, but it will not address the emotion in question. Next time you are suspicious, throw the bot an emotion! A challenging question can also be a way to tell a chatbot from a human on the other end. At any time during the conversation ask about the colour of their clothes or hair. A chatbot doesn’t wear clothes and does not have hair and is likely to answer this question wrongly. A human on the other hand may be taken by surprise and answer this question, or react in a way that you can tell he or she is a human. If you are using an app which enables a to-and-fro conversation, for example a dating app, you can rule out if you are chatting with a bot by asking questions in a creative manner. Alternatively, if you know a less popular language, try and speak with the bot in that language. As bots are usually trained in English or the other popular languages only, they may say they don’t understand you.
Bot-checker tools and websites help track bot activity, especially on platforms like Twitter. Try them out if you think you are interacting with a suspicious account.
Do subscribe to our newsletter and if you have had an extraordinary experience with one, do write to us at firstname.lastname@example.org