New AI Functions as a 'Sarcasm Detector' for Online Comments - Nerdist
NEW
New AI Functions as a ‘Sarcasm Detector’ for Online Comments

Sarcasm is a shady and rather nebulous artform. Sometimes, it’s pretty clear to tell that someone is giving you a helping of snark with their response. The person’s tone and facial expression might be a dead giveaway to the true meaning behind their statement. But sarcasm doesn’t always come across so clearly in written form. It’s even more difficult to gauge for AI software meant to help decipher online comments. But now, there is a program to help detect and decipher sarcastic commentary.

We learned the news at Neatorama. It may seem unnecessary to some people but it does actually serve a purpose. Business owners, particularly those with large volumes of customers, don’t have time (nor the nerve) to sift through thousands of comments.

photo of sarcasm software with sentences showing sarcasm vs genuine text

Ivan Garibay/Ramya Akula/MDPI/Entropy

Two computer scientists, Ivan Garibay and Ramya Akula, secured funding from DARPA to make a program to decipher the true meaning of online comments. The creators’ process is full of intricate details as seen in an Entropy journal publication. It’s a bit much for the average person to understand, but basically, this is a learning module. The program doesn’t appear to have a name yet but it will pick up on certain sarcastic cues and phrases and store those in its long-term memory bank to make future judgments.

So, like most of us, they use technology to make their lives better. However, if the software believes that “this cheap, dull razor is absolutely incredible. Gave me with a lovely skin rash that I’ve always wanted,” is a genuine comment, then that’s not helpful feedback.

photo of sarcasm software with sentences showing sarcasm vs genuine text

Ivan Garibay/Ramya Akula/MDPI/Entropy

This is a pretty tall order because, while there are some general sarcastic phrases, everyone approaches sarcasm differently. And surely they will have to take different slang and colloquialisms into account as well. Either way, we totally think this is a fool-proof way to uncover the truth. What could possibly go wrong?

Trending Topics