I stumbled across a very interesting article about AI (artificial intelligence) lacking the ability to understand and effectively initiate any sort of humor or comedy in a way that even remotely comes close to that which a human can easily do:
What fascinates me the most is level of cluelessness that many comedy/humor researchers and AI developers have when it comes to understanding why a robot can’t simply be “programmed” to be funny in any sort of truly meaningful or advanced way.
The exact same thing can be said of new comedians (and of many who attempt to “teach” comedy as well) who are under the false impression that producing stand-up comedy material that actually works can be obtained by using some sort of “put peg A into slot B” input/out approach.
So in this article I want to discuss exactly why AI will never be able to be “programmed” to be “human level” funny (at least not using the input/output approach that dominates that arena at this point in time).
Then I will tie this information into why most new comedians usually don’t get how comedy actually works and tend to flop on stage — using the same sort of approach used to attempt to program AI to be “funny”.
How Humans Develop Their Sense Of Humor And Comedy Talent
In order to understand why AI can’t be “programmed” to have noteworthy sense of humor or significant comedy talent, one need only look at the process that a human goes through in order to develop a sense of humor/comedy talent.
Required Factors For Effective Use Of Comedy And Humor
On the most basic level, the foundation of a person’s sense of humor (along with the resulting comedy talent they have) involves the continuous collection, accumulation and the communication of reference points (stuff you learn, know, observe, pontificate about, opine about, experience, blah, blah, blah).
What cannot be duplicated when it comes to attempting to “program” a sense of humor into AI are the extensive, lengthy and complex conditions under which a human acquires, understands and ultimately uses the reference points they have at their disposal, specifically:
The Experience Factor. The reference points a person accumulates during their lifetime are much more than mere input/output data — they are gathered as a direct result of extensive experience that starts very early in life.
This experience is gained through countless verbal interactions with others who are also able to express their sense of humor as well.
During this process, humans go through a sort of “built in ” trial and error process where they not only gather their knowledge and reference points, but they also learn from their interactions with others what works to garner laughs and what does not in relation to the reference points being used.
This aspect is unavailable to AI via programming — whatever reference points they have been “programmed” with, regardless of how extensive the associated data may be lacks the trial and error experience of personal interaction needed to not only “connect the dots” between reference points, but to discover and learn what works and what doesn’t when it comes to laughs.
But there is a bit more involved than that…
The Emotion Factor. All the reference points that are the foundation of the human sense of humor are not just “static data” that cannot simply be effectively connected in an A + B = C manner to produce a laughter response. There are emotional underpinnings involved with virtually everything a human experiences, knows, learns and understands.
It is this emotional aspect and the expression of such that provides the human sense of humor context, amplification, subtlety and impact.
A great example of this is the use of sarcasm, which requires certain voice tone, voice inflection and facial expression attributes in addition to the words being used in order to even be recognized as sarcasm.
While there may be a way to program AI to “react” with the appearance of emotion — fear, doubt, joy, pain, wonder, etc., there is no way to connect that sort of programming relative to the associated experiences needed for an appropriate, accurate and genuine sense of humor response.
In other words, there is no algorithm or any amount of programming that can replicate human experience or the wide spectrum of emotions associated with those countless experiences (along with the reference points gathered as a result of that experience).
But wait, there’s more…
The Expressive Factor. Much of what gives a human a sense of humor and resultant comedy talent is NOT just about words they use — it’s a combination of the words and how they are expressed from a visual and auditory manner that communicates a person’s “sense of humor”.
Once again, this is an aspect that is developed on an individual basis over years from childhood to adulthood involving everything I have discussed so far.
The importance of this aspect cannot be understated and can be easily recognized even in one of the most restrictive expressive situations in performing that I can think of which is ventriloquism and the dummy that is used to generate the audience laughs.
If you watch a pro ventriloquist in action, you will note that they are able to take a lifeless doll with limited expressive capabilities and generate huge audience laughs.
But here’s the part that most people don’t see (especially if they don’t know what to look for):
While a ventriloquist’s dummy may have limited expressive capabilities compared to a human, here are the important expressive traits they do have that contribute to the generation of laughter (as controlled by the ventriloquist):
- They have rotational and up/down head movement
- The eyes can usually be opened or closed
- The eyebrows can be moved up and down
- The mouth can be opened to various widths
- There are arm movements as controlled by the ventriloquist.
- The voice tone and inflection provided by the ventriloquist that is congruent with the static facial expression, overall appearance and persona or attitude associated with the dummy.
This is all in addition to the “sense of humor” being expressed by the ventriloquist for the dummy (again, based on what I have already covered).
My point is this — even though a ventriloquist’s dummy may have limited expressive capabilities, those capabilities are absolutely critical to the ability of the ventriloquist to generate laughter when operating that dummy.
Again, there is no algorithm or amount of programming that can connect all these various aspects together to produce a “lifelike” sense of humor in AI.
Now let’s take a look at how what I have presented relates directly to the new comedian and the so-called “experts” who attempt to teach comedy with less than optimal results…
How This Relates To New Comedians
The primary reason why new comedians struggle on stage and fail to get the laughs they want (and could get) can be summed up in a single sentence:
Most new comedians take a superficial, input/output approach when attempting to create and deliver stand-up comedy material without regard to all the other aspects that give them the sense of humor and comedy talent they actually have.
Before I continue please note:
There is no need to study or otherwise deeply understand all the mechanisms of complexity that went into the years long development of one’s sense of humor (and subsequent comedy talent) in order to do well on stage as a comedian.
This “complexity” is built-in to human existence and subsequently is largely taken for granted.
But what a new comedian must understand in order to succeed is that the sense of humor that they have and the resulting comedy talent they express involves far more than just words alone or the order or structure of the words they use to get laughs — offstage or onstage.
Unfortunately, all most new comedians work with is just the words, written in a vacuum using a writing process that is taught and designed for a READER, not a live audience of people who want to experience more than just words alone.
This is the same input/output approach — no different than what is attempted in AI to somehow replicate a human sense of humor — where the experience, emotional and expressive aspects involved are either overlooked, discounted or ignored.
This is why you can’t just study the tire on a car in order to know how to drive that car. You can’t just study a brick to know how an entire house is built.
But to add insult to injury, it’s this same input/output, formulaic approach that is taught by the so-called comedy experts as well.
Just like you can’t “program” the complexity of developing a sense of humor into AI, a person cannot “learn” to have a better sense of humor or comedy talent that they don’t already have.
Unlike AI, comedians do have a choice — which is to use all that makes them funny in a condensed and tightly produced package that can cause audiences to howl with laughter.
But but the new comedian will never reach the performing level they desire using a superficial input/output approach, just like a robot that doesn’t know the difference between funny and Shinola (very old reference but a good one).