The Evolution of Humor
Humor has always been with us. From its roots in prehistory all the way to today’s rapid technological advancements, it’s easy to forget just how innately humor has been ingrained into the human race. When looking at the bigger picture, how has humor evolved? And more importantly, what can humor tell us about the future?
Prehistory
First of all, what is humor? It seems like a simple enough question to answer– in my very first article, I actually used an exact dictionary definition (a funny or amusing quality) and called it a day. However, I’ve come across a more scientific perspective: humor is a complex cognitive function which frequently leads to laughter, a seizure-like activity.
So, how did we develop this cognitive function? And how did we begin reacting with these seizure-like activities? Humor may have evolved with early species of the genus Homo, long before the development of language. At minimum, the emergence of humor in Homo Sapiens may date back 35,000 years. A study from The Royal Society additionally proposed that because laughter triggers the same mechanism that initiates primate social bonding without the disadvantages of grooming, it may have become a necessary addition to group bonding. In particular, involuntary Duchenne laughter, a laughter unique to humans which occurs in relaxed social settings, could be central to group bonding. The involuntary nature of human laughter could indicate an evolutionary component that dates far before other more recent bonding behaviors. Our humorous evolutionary roots may not be observed, but rather heard; both humans and primates share the ability to laugh.
Basic Development
The first humorous conversation was observed in Australian aboriginals. Since then, from Ancient Greek plays, to Shakespeare’s tragicomedies, to American vaudeville and burlesque, to Charlie Chaplin’s slapstick, to late night television, humor has transcended medium after medium. As these new modes of humor developed, so did its interpretations. Notable intellectuals such as Plato, Hobbes, and Freud have expressed their ideas of how and why humor worked. Although many theories of humor have been proposed, there is no universal theory to describe all of its complex aspects. Some more specific theories can additionally be categorized into those listed below, such as the false alarm theory (evolutionary) and the benign violation theory (incongruity). That said, here is a quick rundown of some of the most notable theories:
Incongruity theory: humor consists of two incongruous elements; it is created by contrasting what is normally expected with a violation of social norms.
Relief theory: humor is a healthy way to release repressed aggressive/sexual urges. Unsurprisingly, this theory was conceived by Sigmund Freud, who approached humor using his framework of the unconscious mind.
Superiority theory: humor is a way to elevate social status at the expense of others (boosting feelings of superiority). This theory was first conceived by Thomas Hobbes in his book Leviathan.
Evolutionary theory: humor is an evolutionarily helpful mechanism. The main purpose or evolutionary advantage can range from greater reproductive success, raising one’s own status/socially ostracizing others, providing valuable social information to others, alerting others, to fostering cooperation/social bonding, depending on the specific theory.
Humor and AI
In our current technological landscape, it is difficult not to mention the emergence of A.I. Humor has been regarded as an intrinsic trait among humans– But what happens if A.I. is thrown into the mix?
Some AI researchers are aiming to teach machines to be funny on their own, without a human-programmed setlist. However, the development of a sense of humor in A.I. wields huge implications. For example, Joe Toplyn, head writer for The Late Show With David Letterman and The Tonight Show With Jay Leno, decided to test if AI could learn to make jokes due to the formulaic nature of comedy writing. After all, if comedy can be broken up into formulas, A.I. could potentially create humor by following these formulas. At the International Conference on Computational Creativity, Toplyn presented a paper about Witscript, a joke generation system using a dataset of TV-monologue jokes. Toplyn suggested that if A.I. can understand humor, it can understand how people think and what they like; therefore, the development of more human-like chatbots could make people laugh and feel less lonely. He additionally proposed that A.I. could be used to generate more creative and unbiased ideas than humans.
The recommendation of A.I. as a brainstorming tool was similarly expressed by the authors of a study using entries from the New Yorker’s Cartoon Caption Contest, in which A.I. models and humans were tasked to match a joke to a cartoon, identify a winning caption, and explain why the winning caption was funny. Humans were shown to perform drastically better on all tasks. Especially significant was the gap between human and A.I. explanations of winning captions.
Other researchers doubled down on bridging the gap between man and machine. Researchers at Kyoto University developed an A.I. that could detect and respond to human laughter, with the main goal of emphasizing empathy in conversational A.I. (which in this case, means sharing laughter). Computational humor, the use of computers to generate and understand humor, has the potential to solve the biggest obstacle to A.I: convincing people to create more seamless interactions. Vinith Misra, a computer scientist who worked on A.I. at IBM, Netflix, and Roblox, proposed that humor is just as necessary for interacting with A.I. as it is for human interaction. In order to achieve this, A.I. must be able to emulate the subtle, conversational humor that dominates human conversation. Essentially, we should create A.I with appealing personalities. Still, A.l. has a long way to go in terms of generating palatable material.
In order to appreciate humor, machines require intelligence on par with that of humans. Although there are setbacks, the potential benefits of teaching A.I. humor include having recommendation engines recommend more personalized humorous content, the generation of complex wordplay, improvement in humor writing skill among computational humor algorithms (automatic humor evaluations/suggestions/associations), and the strengthening of the bond between humans and machines. A.I. has gotten stronger in terms of humor generation and detection, but developing an A.I. that can develop/generate all types of humor remains an unsolved issue.
Big thanks to the Streaming Museum for featuring this article! You can check it out at https://www.streamingmuseum.org/luri-lee-evolution-of-humor. As always, thank you for reading, and stay tuned for next month!
Sources
Comedy -- Britannica School. school.eb.com/levels/high/article/comedy/110456.
Communications, Frontiers Science, and Frontiers Science Communications. “Sharing a Laugh: Scientists Teach a Robot When to Have a Sense of Humor - Science and Research News | Frontiers.” Science & Research News | Frontiers - the Latest Science & Research News From Frontiers - Publishing the Most Cited Open-access Journals in Their Fields, 14 Apr. 2023, blog.frontiersin.org/2022/09/15/can-a-robot-laugh-ai.
Delahoussaye, James. “How Can Our Relationships With Computers Be Funnier and Friendlier?” NPR, 3 June 2022, www.npr.org/transcripts/1102642165.
Dunbar, R. I. M. “Laughter and Its Role in the Evolution of Human Social Bonding.” Philosophical Transactions of the Royal Society B, vol. 377, no. 1863, Sept. 2022, https://doi.org/10.1098/rstb.2021.0176.
“It’s No Joke: AI Beats Humans at Making You Laugh.” HBS Working Knowledge, 20 Nov. 2019, hbswk.hbs.edu/item/it-s-no-joke-ai-beats-humans-at-making-you-laugh.
Purtill, Corinne. “Artificial Intelligence Can Now Craft Original Jokes—And That’s No Laughing Matter.” Time, 4 Jan. 2022, time.com/6132544/artificial-intelligence-humor.
“That’s Funny – but AI Models Don’t Get the Joke | Cornell Chronicle.” Cornell Chronicle, 26 July 2023, news.cornell.edu/stories/2023/07/thats-funny-ai-models-dont-get-joke.
Winters, Thomas. “Computers Learning Humor Is No Joke.” Harvard Data Science Review, Apr. 2021, https://doi.org/10.1162/99608f92.f13a2337.
