AI intelligence has taken massive strides in the past year. It is now possible for face and object recognition, understand different voices, and many more complex variables.
One thing that AI can't seem to master, is humor. Since different people find things funny, it is impossible for AI to predict humor. Some psychologist think that all good jokes share the same qualities, so AI may be possible.
In this article, Radev took 300,000 captions written for 50 New Yorker cartoons since 2005. He linked different captions and gave them rating based on the topic used. He then tried to figure out what topics linked to most liked captions.
He found out that the funniest captions were based on negative sentiment, human-centerdness, and lexical centrality.
Radev tried to imagine an AI system that could select the funniest captions based on these findings. He concluded that it would be virtually impossible to create an effective machine to pick out humor.
In today's day humor is not something AI can distinguish, but in five years, it might be a different story.
https://www.technologyreview.com/s/539291/an-algorithmic-sense-of-humor-not-yet/
No comments:
Post a Comment