Microsoft Is Teaching Robots The Art Of Humor
Everyone finds Robots funny; they have always played the straight man oblivious to any joke played at their expense. Now The New Yorker and Microsoft are attempting to teach a robot to be intentionally funny, so that it can help with its caption contest according to Bloomberg. Since its introduction in 2005, the caption contest has become a firm favorite with New Yorker readers. Readers have to put a caption to a black and white cartoon and the readers send in their best efforts to finish it. It has proven to be too popular and now its editor – Bob Mankoff is being deluged with over 5,000 contest entries each week. This has been difficult on his assistants.
"The process of looking at 5,000 caption entries a week usually destroys their mind in about two years, and then I get a new one," Mankoff told Bloomberg.
They had to find a way of going through all those entries without becoming completely bored of the humor. Microsoft's solution was to use a robot, so their researchers teamed up with The New Yorker to try and create a robot capable of identifying which of the captions were funny.
The researchers fed The New Yorker cartoon and caption data to the robot, attempting to teach it how to differentiate between jokes and similar humor . Although the ''top captions' list didn't align completely with the robot's, all the editors favorites were in the top 55.8% of the robots selections according to Bloomberg.
The robot couldn't identify the best and funniest caption, but it was able to discard the majority of truly bad ones. That alone could save at least fifty per-cent the workload of Mankoff's assistants.
