3/8/2025–|Last update: 16:32 (Mecca time)
In the past years, the uses of artificial intelligence have expanded significantly to storm many fields, starting with scientific fields such as programming and other different engineering sciences, and even humanities such as writing stories and texts in general.
By virtue of his work in comedy with senior program providers such as David Litrman and Jay Lino and his presentation of many courses in writing comedy, prominent comedian Joe Toplin asked about the capabilities of artificial intelligence in writing comedy and jokes.
The result of this question came in the form of a sarcastic artificial intelligence tool called “Witscript”, and it is available in the form of a web application that can be used for a small monthly subscription, and works to generate jokes and comic comments through the addresses of the news or texts and pictures that it provides.
Toplin was also able to test the capabilities of the tool by entering into a three -day comic competition to produce a group of different jokes, then measure its quality and the audience interacted with it.
Measuring the ability of artificial intelligence to generate comedy
The competition that Toplin entered with its artificial intelligence tool included generating a group of jokes on eight permanent news topics, then presented to a living audience in Los Angeles and measuring their reaction to these jokes without knowing their author.
Toplin cooperated with the nerve scientist and comedy Uri Amir to measure the reaction of the public on these jokes, and the test found that the impact of the jokes of artificial intelligence and human jokes is very close to influencing and laughing, according to what was stated in the report of “Smithsonian” magazine.
In another experience, social psychology researcher Drew Gurines of the University of Southern California and his colleagues found that the “Chat GBT 3.5” model was able to generate texts in the distinctive satirical style of The Onion magazine, after providing it with more than 50 titles of the magazine.
These experiments raise a real question about the ability of artificial intelligence to understand the sense of human humor and the possibility of imitating them using deep language models, especially since a large part of the sense of humor is related to societal awareness and the context in which jokes appear.
The researcher, Tristan Miller, believes that learning artificial intelligence models to produce humor is a necessary part of its journey to simulate human experience and use the language in a manner that resembles humans, and that humor is the most human and complicated aspect of the language from his point of view, according to what was stated in the magazine.
Reservations on the sensitivity of artificial intelligence
Many scientists and experts see a set of concerns and reservations about the ability of artificial intelligence in the production of humor, specifically when it comes to biasing artificial intelligence and its sensitivity to human patterns clearly, according to the magazine. Artificial intelligence can recycle bad stereotypes in some types of humorous content online, thinking that they are appropriate and have no problems with them.
The Smithsonian magazine report indicates that the Topblin and Amir experience relied on a group of jokes that Toplin – a professional comedy – chose it among all the jokes that the Watt Skript tool wrote, which shows that the human evaluation is still necessary to determine whether the jokes are already funny.
Gorens explains that the “Chat GBT” model was not designed as a basis for writing jokes, but to create and expect usual texts, and therefore he is unable to feel emotions associated with laughter and estimate whether the joke is appropriate or not.
Despite this, the model can generate funny content sometimes, which causes Gorens wondering: Is it sufficient for a large number of data and the ability to identify patterns to produce funny content without the need for a personal estimate?
Christian Hemelman, a researcher in computer linguistics at the University of East Texas, believes that the real humor is related to the intentions carried by the owner of the joke, as humor allows you to play with meanings as you wish, so the most important component in the composition of funny jokes can only be presented to humans, which is emotional interaction with humor.
But with the rapid developments that occur in the artificial intelligence sector, can we see a model able to generate very funny jokes and attach them with cinematic scenes or sounds that add a deeper context?