This new AI program acts as a sarcasm detector for comments online

Not everyone understands sarcasm and not everyone can dish it out. Sometimes it is pretty clear when someone is being sarcastic if you pay attention to their tone and facial expressions. However, that does not apply in written form when you cannot see the person. And it is also difficult for AI (artificial intelligence) software to gauge sarcasm as it deciphers online comments. But that’s set to change. There is a new AI program that can help detect and decipher sarcastic commentary on online posts. 

First reported by Neatorama, this new AI tool might seem superfluous, but it isn’t, it actually serves a purpose. Particularly for business owners, those with a lot of customers, who do not have the time to go through thousands of comments, a program like this will help them get the feedback they should be paying attention to - like a disgruntled, sarcastic customer. 

Created by two scientists, Ivan Garibay and Ramya Akula, who secured funding from DARPA, this program works on trying to decipher the true meaning of online comments. As the article in Entropy journal explains, the process is full of intricate details and quite a lot for a layperson to understand. Basically, it is a learning module. The unnamed program picks up on certain sarcastic cues and phrases and stores them in its long-term memory bank, like all other AI programs, and this is what helps it make future judgments.

It sounds helpful, truly. But it remains to be seen if this program can actually weed through enough content and can learn it right to point out that - “this cheap, dull razor is absolutely incredible. Gave me with a lovely skin rash that I’ve always wanted” is a sarcastic comment and not genuine feedback, as Nerdist points out, only then is it going to be of any help.

Given that there are only a few general sarcastic phrases and almost everyone approaches sarcasm differently, it’s a lot for the program to learn. Then there is slang and colloquialism that has to be taken into account as well. Guess there’s only one way to find out, right?