March 28, 2019 • 41 mins
Why is it so hilarious when robots and artificial intelligence fail? What does it reveal about comedy itself and our technological anxiety? Robert Lamb and Joe McCormick explore in this Stuff to Blow Your Mind two-parter. In this second episode, the joke's on us.
Learn more about your ad-choices at https://www.iheartpodcastnetwork.com
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:03):
Welcome to Stuff to Blow Your Mind from how Stuff
Works dot com. Hey, welcome to Stuff to Blow your Mind.
My name is Robert Lamb and I'm Joe McCormick, and
we're back with part two of our discussion of the
plate a sex Mac and I think that's what we're
gonna call it, the comedy of errors that exist in
(00:25):
the machine and artificial intelligence world. That's right. We were
gonna do it as one episode, but it just went
too long, had to break it up into two. But
it has everything it has, uh, it has humor, it
has theories of humor, it has killer robots, references to RoboCop,
everything you could ask for. One of my favorite things
explaining a joke to death, yes and then yes yeah. Yeah.
(00:49):
Basically this is a lot of joke explaining. But I
think by the end of it, everyone will still retain
their ability to laugh. All right, So, in thinking about
what's so funny about machine failure, uh, and and even
machine success along a certain you know, parameter of distance
from what from what it's ultimate level of success would be,
you know, AI, that's sort of succeeding but still isn't
(01:11):
really there. Um, I guess it's worthwhile to look at
what are some of the main theories out there about
what humor is? Like, why do we find anything funny
not just machines? Yeah, because it becomes increasingly difficult to
really uh nail it down, you know, you come back
to that whole idea of it's funny because it's funny
or you know what when you see it. Uh. This
(01:33):
topic has come up a few times on the show
before um most recently two thousands, seventeens laughing during horror movies.
That was when Christian was on the show and we
talked about you know, what is it about horror movies
like frightening, terrifying films that that elicits laughter from us?
And then there were a few other ones in the past,
The healing power of laughter, the killing joke, funny or Die.
(01:54):
These are all episodes of stuff to blow your mind.
You can you can check out in the vault if
you so desire. Yeah, and so what if you go
and check those out, I'm sure you as you'll probably
found before. There's no one theory of human right and
nobody agrees, but there's sort of a host of competing
theories that I would say aren't even always totally contradicting
(02:16):
of one another. They they sort of partially overlap and
partially contradict one another. Right, Yeah, there's not really a
theory of everything per se, but like each one feels
a little bit right, like each theory is is kind
of fondling the elephant. Yeah, exactly. Uh So, a brief
run through of a few at the top. And there's
no way to explore all the theories of humor. But
(02:37):
if you are the most commonly cited and most popular ones,
one would be the superiority theory. And this is propounded
in some classical works, like in the works of Plato
and Aristotle. This one is actually is kind of nasty,
But sometimes I guess humor is kind of nasty. It
proposes that we laugh when we notice someone is less
fortunate than us in one way or another, and the
(02:58):
laughter comes from our feeling that we occupy a place
of superiority. We sort of deploy it as a form
of scorn on somebody else. And Plato and his Philibus dialogue,
which is discussing the nature of pleasure and different kinds
of pleasure, like why pleasures of the mind might be
superior to pleasures of the flesh. Plato has Socrates claim
(03:20):
that we laugh at people who do not recognize their
own misfortune, for example, when people are stupid but think
themselves brilliant, or when people are ugly but think themselves
very handsome. And I would say this clearly doesn't cover
all types of humor, Like, you know, what about self
deprecating humor when you can laugh at yourself, But there's
(03:40):
something there that is clearly present in some humor. Yeah, well,
you know, when it comes to self deprecating humor and
laughing at yourself, I think it does make sense when
you consider our abilities to sometimes step outside of ourselves
and see ourselves as a character in a story, sort
of the infectious nature of narrative there. But then also
in the way that we use theory of mind way
(04:01):
too much to try and imagine how others see us.
So maybe, like, if you're trying to roll with this theory,
you'd say that self deprecating humor is almost when we
pretend that we are ourselves someone else and you like
step step back and you scorn a different version of yourself. Yeah,
basically like or you're just kind of imagining how scornful
other people were being to you because of something you
(04:24):
did or said or thought. Yeah. Um, I mean another
thing I would just say is that unless you're a
horrible person, and maybe you know, Plato and Aristotle were
in some ways very smart but also very horrible people,
unless you're a really bad person, you don't find most
types of misfortunes of other people funny. I mean, I
do see a certain distinction along Plato's lens in the Philippus,
(04:46):
where like a person just failing to do things that
you could succeed at is generally not very funny. Like
somebody trying to solve a math problem you could solve
and they can't do it, that's not funny. But like
if a man proclaims himself to be one of the
greatest brains of all time and then is making basic
errors of logic and math, that's usually funny. Yeah. I
(05:08):
think a lot of it hinges on just you know,
what are the stakes and what sort of misfortune is
taking place? Like Another sort of example of this is
when someone thinks they are refined about something, uh, and
maybe they're a little snobbish about it but there, but
it turns out their tastes are not quite as refined
as they thought. Um. I think there was an episode
of Three's Company back you know in the day, in
(05:31):
which a don Knots character is ordering like a fine
scotch at a bar and he's really talking it up
like he knows what he's talking about, exactly what kind
of scotch he wants, and then when the bartender finally
serves it to him, he requests that the rest of
the glass be filled up with root beer, and you know,
it's like, like, that's that's so that's hilarious because he
turns out he's not as as refined as he thought.
But then it's also like we were being snobbish in
(05:54):
making that judgment. Uh, And it's and that's always kind
of a dick move, right because in anything like whatever
your thing is, that you feel like you're refined and
be at food or drinks or film or literature, like
there was a point in your life when you were
not when you were the person that was the subject
of scorn. And you know, ultimately we're being fair. You
should celebrate that they are at all interested in the
(06:15):
in this thing at all and that they are on
hopefully on some sort of the same journey that you are,
but instead we often laugh and said, can you believe
that guy just put root beer all in that sky?
I think that's a really good point. So I don't know.
I would say I definitely see something going on here
that there are definitely elements of this in humor, but
(06:37):
I don't think it explains all humor. I don't know
if maybe, maybe, maybe not. I don't know if it's
a good fit for why machines are funny, because obviously
machines aren't people exactly, but maybe we sort of see
them as people when we think they're funny. Yeah, I
think if we if you factor in the actual and
or perceived abilities of AI and robots, either either what
(06:58):
they can actually do or more I think, to the
point what our science fiction and our futurist scenarios have
proposed that they will do or could do one day,
because we end up with this situation where like we
watch our room but do something stupid, and there's it's
hilarious because like, ha, you're never going to terminate or me,
like you can't even you can't even deal with the
(07:20):
uh with this rug or this pile of cat vomit. Well, actually,
though I would say another reason this does kind of
connect is it does kind of connect to Plato's thing
about laughing atself ignorance, because one of the funniest things
that I sort of identified earlier about like the D
two O nine scene and all that, one of the
funniest things about the ways that machines fail is the
(07:41):
way that they are just so completely oblivious to their
own shortcomings, the way they just uh, the way I
think of it is that they fail and then plow
straight ahead. You know, they have they have done something
crucially wrong, and they do not appreciate this fact. All right,
let's keep rolling. What are some other theories can discuss? Okay,
next one would be like the nervous energy release theory.
(08:04):
And this is not surprisingly at all a take that
was popular with Freudians. Under this model, laughter is sort
of a pleasurable release of built up psychic tension caused
by fear or apprehension. Uh. And there's a certain kind
of logic to this, right in the way that's a
good Comedians repeatedly build at tension and then dissipated in
various ways. I bet this also goes with the laughing
(08:26):
during horror movies thing, you know, Yeah, for sure, Like you,
you get your tension built up when there's something scary
is about to happen, and then after it's over you
you release the tension and you can laugh. Yeah. Certainly
with with comedians who really leaned into like the awkwardness
and the tension, like say like an Andy Kaufman. You
know where their whole their whole bits or if you
(08:46):
want to call them bits, where it's really just making
everyone uncomfortable. It's not even about punchlines. It's about just
creating this this tension that at some point has to
be relief. Like at some point all you can do
is laugh. Yeah, So there's something there. I I this
clearly doesn't account for everything that's funny, you know, it
doesn't account for puns and all that kind of thing.
Uh It, I don't know if I can see a
(09:07):
way that it really fits with why technology failing is
all that funny? No, Yeah, I don't think. I don't
think there's anything particularly insightful about this theory regarding what
we're talking about here today. Okay, here's another one. How
about the adaptation, uh to signal play so that laughter
would be an evolutionary adaptation that occurs when we're engaged
in pretend aggression during play behaviors. It occurs as a
(09:30):
signal to help everybody involved understand the fact the behaviors
like chasing and wrestling and stuff are meant as play
rather than as genuine aggression or harm. Yeah, this one
I think is important to mention here, if only for
the reason that it touches on the social nature of laughter,
that that laughter is also about communicating with other people
(09:51):
regarding various threats, etcetera. Um. It's also I wonder if
this is one where our storytelling has kind of kind
of led us astray as well, though, because how many
um individuals that actually intends harm are laughing hysterically or sardonically.
Like I'm sure it happens, I mean, and maybe one
(10:14):
or two scenarios even come to mind where someone was
did something horrible and violent and it was seen laughing,
But I don't feel like it probably occurs anywhere as
near as often as it does in movies where villains
are constantly laughing. Be it, you know, an extreme example
like the Joker, or or just any you know, you know,
sardonic villain who just doing maniacally laughs while attempting to
(10:37):
destroy the world. Uh yeah, I mean I would say
this probably does go along with the theory then. I mean,
if it if somebody is laughing while they're actually attempting harm,
there you know their victim is not laughing so like,
so yeah, I think this goes along there. Like and
if they are, then it's like that scene in the
original um what was a Little Shop of Horrors where
(10:58):
or maybe it was in the remake is well too,
where the uh, the the the the the evil dentist
is going to work on this person and this person
just wants it, They want the pain, and it just
totally messes up his his vibe. Yeah, I haven't thought
how it would accommodate the idea of like villains laughing
while they do what they do. But I mean, I
think this is clearly true in some cases that people
(11:18):
laugh during mock aggressive play behaviors. That dogs wag their
tails during mock aggressive play behaviors. It's like, you know,
they're growling and snarling and tumbling, but you can tell
they're not really fighting because their tails are wagging. And
this is a nice lead into one of one of
my favorite theories, and this is one that we discussed
on the show before in the past, and that's the
(11:39):
benign violation theory. Yeah, and this theory is very popular now.
This theory proposes that humor occurs when one we perceive
a way in which rules or norms of some kind
are broken. And these rules can be any, you know,
all kinds of things. They can be grammatical rules, they
could be biological classifications, they can be the rules of
(11:59):
making sense, they can be moral norms. Uh. And then second,
we recognize that there has been a violation, but it's
not harmful, it's benign in nature. And then third we
see this contradiction between the fact that there's a violation
and the fact that it's harmless, and that contradiction triggers humor.
Like we said, this one is very popular. It's supported
(12:20):
by some empirical research. Just one example of research and
support that I came across as a two thousand ten
study in Psychological Science by McGraw and Warren called benign
violations making im moral behavior funny and too so. To
just cite one example that they talk about within that study, quote,
people who are more weakly committed to a norm can
(12:41):
recognize the violation, but are less likely to be threatened
or to directly experience the violation's repercussions. Consider a news
story about a church that raffles off a hummer suv
as part of a promotion for its members. Engaging in
such a secular promotion jeopardizes the sanctity of the church.
End Although most people consider churches sacred, churchgoers should be
(13:04):
more strongly committed to this belief than are people who
do not attend church. So the researchers looked at this
and they found that while churchgoers and non churchgoers alike
were about equally disgusted with this story, you know, the
idea that a church would be trying to get people
to join by giving them a chance to win a car,
non churchgoers more often found the story humorous, and so
(13:27):
that was like non churchgoers to churchgoers finding it humorous
was nine percent to sixty two percent. So the idea
is that non churchgoers saw this as a benign violation
and thus funny, and churchgoers saw it as a real
harmful violation and thus not funny. Yeah, and I think
that's something that's important to keep in mind. And you know,
(13:47):
when when looking at the benign violation theory is that
how ultimately subjective it is going to be. And of
course that is comedy as well, so it matches up
in that regard. Right, It depends on whether you actually
see something as a violation and whether you actually see
something as harmful or not. Now, there are different ways
you could play around with the definitions and of this
theory to like, like, you could make it very elastic
(14:10):
to accommodate a lot of stuff, or you could more
narrowly tailor it. I would say one problem potentially with
it is that lots of things that are benign violations
aren't funny at all. Uh. An example I came up
with is, imagine you are not supposed to take home
office supplies, but you do you take home a pin. Well,
that's a violation of the rules. It doesn't it's not
(14:32):
really harmful, it's benign, and it's not really funny. Right.
But on the other hand, I would I think the
counter argument here would be, okay, taking one pin home
from work, like that's I do that all the time,
most of us do. It's one thing to take a
pin home, even accidental, accidentally, take one home to you know,
purposely take one home. But what have we upped it?
What have instead of accidentally taking a pin home from work,
(14:55):
you accidentally took all the pins home from work. That's
funny now that it was getting into you know a
little more. And I'm not saying that there's it's the
funniest gag of all time, but certainly there's more room
for hilarity in this scenario. Well, that starts to be
more like a machine type comedic mistake because it's so uh,
it's so over the top. It seems to kind of
(15:16):
like not just violate a norm but just plows straight
through it and goes all the way down right where.
I guess another way that would work is if one
meant to steal something of greater value, like one intended
to to um to perpetrate some greater violation of office norms,
and then in the end you only stole a pin,
(15:36):
but even in a done land as well, and it
really needs to be a box of pins or all
the pins. Yeah, I guess you could say. I mean,
I think you could probably play around with exactly how
you define the terms of benign violation to maybe accommodate
those counter examples. I would say also though lots of
things that are funny too, people can feel like they're
not at all benign, and like think of videos of
people falling off skateboards and smashing their faces or the
(15:59):
crotches and the handrails and brick walls and stuff. People
find these things quite funny, and people can get really
hurt in them. Yeah, I think with with that, I
think part of it is that we generally see confirmation
that they survived, or if we don't see confirmation that
they survived whatever, they're horrible, you know, skateboarding accident was,
then we at least don't see them going to the
(16:20):
e er, you know, like we have no additional information
to go on, no additional context, and we can just
kind of assume that they were okay. So, and I
also was wondering about this. I wondered what extent comedic
interpretations of slapstick injuries in which harm generally doesn't extend
beyond the punch line. Um, if these program us to
(16:41):
to make these kind of comedic interpretations. Uh, for instance,
you know, when you think about comedic violence, you think
about the Three Stooges, and certainly they're just always brutalizing
each other and then they're fine, and then they're back
for another adventure. They're fine. You don't worry about their health.
Take a good old fashioned three Stooges fingerpoke. Um, you
(17:02):
know we have in the in the eye do you
have two fingers and then you go right for the
eyes and if nobody gets the block up, bam, there's
like as more of a blank noise, right. So yeah,
so when the Stooges do it, uh, it's hilarious, right,
there's a lot of slapstick hilarity with with the double fingerpoke.
But and then likewise you look at something like pro wrestling,
(17:24):
in pro wrestling, which of course is this like sports
theater thing. Uh, the fingerpoke is a comedy spot, you know,
it's it's if someone goes for the double eye poke,
it's going to be a moment of hilarity as well.
But if somebody does it in a basketball game, that's
not really funny, right, or or if someone does it
in most if not all combat sports, it is it
(17:47):
is a band maneuver. Like generally, if you're going for
somebody's eyes, like that's either it's either really evil or
it's a move of desperation. Yeah, this does certainly make
that to whatever extent the benign violation theory is true,
it's highly context dependent and and maybe the truth there,
if there is something to this theory is that um
(18:07):
is that the context makes us determine what's benign and
what's not right, and it's gonna be highly dependent on
the culture you're in, the time you live in. I mean, certainly,
just thinking of workplace humor, there are a lot of
violations today that are serious that may have been perceived
as more benign in days past, or as or seen
(18:27):
as not violations at all. Oh yeah, this go back
to like the nineteen eighties, and like workplace sexual harassment
was a common like sitcom gag at the time. It's
just like, oh, it's hilarious, you know, he's he's harassing
Tina again. Uh, And now like that's not really funny
at all anymore because we see that, I think, I
think because people see that as a more genuine violation,
(18:49):
not as something that's just a harmless you can wave
it off, that's just ted being ted. So how do
you feel about about benign violation and neural networks? Well, strangely,
I do see part of a fit here. Um, because
I along the lines of the Church suv example. The
more serious and reverent the context in which the technology
(19:10):
is presented, the funnier it is when it fails. I
think about like ED two oh nine. This the scene
is sort of made by the fact that Dick Jones
and all the biz bros Are proudly boasting about this
marvelous new technology that will be the future, and then
it fails in such a spectacular way. I think in general,
the failures of AI are much funnier when they're presented
(19:32):
within a context of us knowing about people who believe
in AI messiahs and you know AI Basilisks and Satan's
and all that who who uh see this great future
of of power and dominance among Aiyes, watching them mess
up in hilarious ways becomes all the more poignant and perfect.
All right, time to take a quick break. Well, we
(19:54):
will be right back. Thank alright, we're back. So one
more theory is the in congruity resolution theory. This is
a theory that's had various forms over the years, for
example advocated by Manuel Kant and others, but essentially the
idea here is that there is an incongruity between expectations
and reality. Laughter occurs when we realize this incongruity can
(20:17):
be resolved, and this is there's clearly a strong element
of this in lots of humors. Again, what joke involves
a set up and then a punch line, and most
often the set up sets you up to expect one
type of thing. The punch line subverts your expectations. So
that's clearly there. But then again, there are lots of
things that are surprising and turn our expectations on their heads,
(20:39):
and they're not funny. Yeah, like we incongruity has to
be novel. Yeah, you know, it can't be a situation
where I thought I was buying mozzarella, but I really
got monster. That's not that. That's really not funny. Now,
it would be funny if you thought you were buying
mozzarella you got home and it was like rat eggs,
especially because rats don't lay eggs. Yeah, it's like that. Yeah,
(21:00):
that would be that would be a lot more of hilarious.
One different but somewhat similar theory. I was reading about that.
I'd like to read this book now that I know
it exists. It's a two thousand eleven book from M
I t press called Inside Jokes by the author's Matthew Hurley,
Daniel Dennett, and Reginald Adams Reginald Adams Jr. And they
(21:21):
argue against the sufficiency of any of these models we've
talked about already. They propose instead that humor is an
evolutionary reward system. It's a pleasurable reward feeling that we
get when we recognize the inappropriateness of a mental representation
and then fix it. So it's the brains. It's it's
an internal reward system like we get from other things
(21:43):
that are good for the body, like eating or something. Uh.
It's the brain's reward system for debugging itself. And so
you can imagine this is why many jokes have a
set up. The setup prepares you to establish an inappropriate
or incongruous mental representation. The punch line suddenly creates that
inappropriate representation, and then the debugging that immediately happens in
(22:06):
your brain following is rewarded with this pleasurable feeling of humor. Interesting.
I think that's something that that probably matches as a
model that matches with a lot of our experience of
of having hilarious or ridiculous notions enter our head and
then you just miss them, or you even don't, you're
not even considering them, but they just rise up from
(22:27):
the depths of your psyche. Yeah, and I think it
kind of fits well with what's funny about each entry
in these like machine generated pieces of text, because each
moment you're going through this text, you're sort of being
set up to think about the text in a certain
way given the category you're already looking for, like names
of spells and stuff. Then constantly you're given something that
(22:49):
doesn't really fit, and then your brain has to debug
why it doesn't really fit, and that feels good. I'm
not sure it's a perfect theory that covers all of
humor either, but I am interested in Maybe you know,
another thing is that, like you said earlier, it's possible
there's no single, perfect, overarching theory of humor, Like what
if humor is a category held together not by a
(23:10):
single all inclusive set of criteria, but by a Wittgensteinian
kind of family resemblances principle. There's no one thing that
that all humor is. Humor is instead multiple different related
sets of things, none of which are entirely contained in
a single box. By the way, you know, if you
open that box, it's going to be one of those
(23:30):
spring snakes that comes flying out, Like open the fake
cannon nuts. Which one does that fit in? Um? I
guess you know. It has to do with expectation, but
also it's benign violation. So if you're expecting nuts, not snakes,
but you've got a snake. But then, well it's not
a real snake, which is a perfect animal to include
in that gag given our you know, our our evolutionary
(23:51):
response to a serpentine form. What if it's a real
snake that bites you on the throat, I would like
to see that happen more or what. One of my
favorite gags that I did this is that a workplace
at some point there was this I had to like
kind of this lame boss who had a can. He
had this exact gag of fake nuts, fake peanuts, and
if you opened it, a snake jumped out. Um. So
(24:14):
one day he was gone for the day and so
I took the snake, got rid of the snake, and
I filled it with actual peanuts, and it gave me
so much pleasure. I don't I never even got to
witness the payoff. But I just love the idea of
him going to like pull this out on somebody or
to just amuse himself, and then it's actually filled with nuts.
And I don't know how lame the boss was. This
was This is kind of a time in my life
(24:35):
when all bosses were lame. So if any of my
previous bosses are listening to me, I'm probably not talking
about you. Well, there's another theory of humor that kind
of fits that because of the way in which humor
sort of interlocks with our expectations of people based on
their individual characteristics. Uh So, the next one clearly doesn't
account for everything that's funny, but it does have a
(24:56):
kind of elegant, elegant connection to what we're talking about.
So this is a theory of humor put forward by
a philosopher named Henri Bergson in a book called Laughter,
published in nineteen hundred and One of the ideas Bergson
puts forward here is that laughter is often our response
to rigidity, inflexibility, and failure to adapt. And this is
(25:17):
one of the reasons why comedies involve us being aware
of like the particular flaws of a character and then
seeing those same flaws acted out over and over again.
Think about sitcoms, like, it's not just that the characters
have failures, but they tend to fail repeatedly in in
individually characteristic ways. Homer Simpson is lazy, and he's thoughtless,
(25:40):
and he's motivated by donuts and gluttony and all that. Uh.
Michael Scott in the office is needy and attention seeking
and unaware of how he's perceived by others. Characters have
these sort of trademarks and they keep making the same
kinds of mistakes over and over, and it's their inability
to adapt that makes humorous in this point of view. Now,
(26:01):
clearly this doesn't account for everything that's funny, but it
sort of ties in loosely with the Hurley Dinnett Adams
theory that humor rewards adaptation and mental debugging. These people,
if they're making the same mistakes over and over, they're
not debugging, and you're like debugging by watching them. But
then again, of course we have to come back to
sort of the benign violation theory a little bit here,
because we don't want characters that uh failed too much,
(26:25):
you know, like you don't want it to be like
a tragic level of failure, right yeah. And of course
in reality people fail to adapt all the all the time,
and ways that aren't funny but are actually like sad
or tragic or you know, it's something very bad about it.
And then there's that whole sliding scale too, because especially
with something like like Michael Scott, especially like the Office
got into so much awkward territory just every week, week
(26:47):
after week, and um, a lot of that really got
into sort of painful, pitiful territory, so that they definitely
changed exactly where the line was in the sand, but
but still ultimately, like you know, Michael Scott didn't die
at the end of an episode, well, it helped by
coming back and making the character lovable, like he wasn't
(27:09):
getting into that like sad awkwardness and then just ending
in a place where you hate him, right yeah. But anyway,
I think this theory about like rigidity and inflexibility and
like repeating the same types of mistakes without adapting has
some strong purchase on the funniest features of machine failure comedy.
It's not just that the machine fails, but that the
(27:31):
machine fails repeatedly mechanically. According to inflexible guidelines, like the
way it just keeps following the rules, when a neural
net text generator spits out not just one nonsensical phrase,
but like reams and reams of text without acknowledging or
knowing it all that the first few words didn't make
any sense. Or when I think about when a glitching
(27:52):
video game character is stuck inside a table and just
keeps running in place and talking. That's one of my favorites.
Or that lack of contacts, lack of context, lack of
self awareness, and lack of adaptability. The way machines just
plow ahead despite problems that are obvious to us that
they're clearly not aware of. I think that's key to
(28:14):
what makes them so especially funny. And in this sense,
they're kind of like the Homer Simpsons and the Michael Scott's,
the characters with inflexible, repetitive behaviors that lead to disaster
but which they never learned from or change. You know,
I can't help but come back to the idea of
of a child and all of this and the child
analogy here, so you know we can we can laugh
(28:34):
at our own children's mistakes are weird interpretations because we
know that they're they're largely going to adapt, they're going
to grow up, uh, and that they're going to leave
behind their their dragons or hopefully they're just gonna remember
how to summon them in appropriate ways now. But if
they're the robot is hilarious because it continues to fail,
and of course also because it is a machine and
(28:56):
not a human child. Perhaps part of the issue too,
is that a machine is but a prototype, and it's
next phase of advancement does not necessarily exist within itself,
but in the next prototype and another individual, if you will.
I think that's a really good point. Yes, Like, so
a child can make a funny mistake, but then the
(29:16):
child immediately within themselves learn something and or you know,
at least as the potential learn something and grow and
get better at not making that kind of mistake anymore,
which in some cases kind of tragic because the mistakes
are glorious. Oh yeah, like half the in jokes that
my wife and I make are two things that my
my son has not said in years, but he said
(29:38):
it was and it was hilarious, and now we repeat
it to each other. But like most of these machines,
even though they might have the ability to learn from
their mistakes. They learn from their their mistakes in like
new instantiations of one of themselves, like you're saying, it's
the next prototype. They don't automatically learn from their mistakes.
You have to do something to them to reboot them
(29:59):
or retrain the them or something right. And if it's
a video game that, yeah, this, this character is gonna
keep getting stuck in that table. Maybe there'll be a
patch that fixes it, but for the most part, it's
going to be the sequel that tries to get it right.
All right, we're gonna take a quick break, but we'll
be right back. And we're back now. I think this
is all one of one of the reasons that the
(30:19):
Black Mirror episode Metal Head from is so effective. Oh
it's it's it's one of my favorites. Um this was
This was again seen. And this came after years of
us watching impressive, yet ultimately clumsy and imperfect Boston Dynamics
walking robot videos, videos of such you know entities as
(30:42):
Big Dog, you know, which kind of has this springy
brant kind of a sound as it prances along, and
it's very impressive to look at and yet at the
same time, you know their videos of being kicked around,
and you know, certainly if it's succeeding a lot, but
also not quite it's not reaching that level of like
mature technology. Yeah, though I worry how much I already
(31:05):
sympathize with it, Like watching those Boston Dynamics robots when
somebody kicks the dog. I know it's just a robot,
but I don't like that person who kicked it. I'm like,
that's cruel. How could you do that? Well? In the
Metal Head, Uh, Basically, the whole thing is kind of
a post um apocalyptic scenario, killer robot scenario, and the
(31:25):
killer robot we see is not a juvenile prototype, but
what you might classify as an adult that the technology
in a proceed at a perceived level of maturity. So
the Boston Dynamics Big Dog all grown up into a
terrifying lethal, quadrupedal killing machine that it overcomes every challenge
(31:48):
in the natural or man made world that it is
presented with. Uh. And I think that's that's one of
the reasons it is terrifying, because it's kind of like, oh, yeah,
these things are going to grow up. We're for a
lot of money and effort into making them grow up,
and uh and and how are we going to relate
to that technology when it reaches that point? Well, that
brings us back to something you were saying earlier about
(32:09):
like the CS Lewis quote, Like clearly our machines and
their failures I think are becoming much funnier over time.
Like a steam engine designed to pump water out of
a coal mine is really not funny when it malfunctions,
maybe like cut somebody's arm off, like it's it's not hilarious.
Automatic doors are still not very funny. Maybe a little
(32:30):
bit funnier, like I was saying, when they're trying to
close on something over and over, and you know when
it gets repetitive like that room buzz are funnier. Anything
attempting to generate real human language or simulate direct human behaviors,
like say the DARPA challenge bipedal robots. You know when
you watch them like collapsing in the sand, or like
(32:50):
when one of them like opens a door and the
door goes open and then they just crumple in a
heap on the ground and that kind of thing. It
is definitely even funny to hear, like machines seem to
get funnier. The closer they get too resembling real human behaviors,
the better they simulate. And I wonder, so it's almost
like there's an uncanny Valley of humor in a way. Yeah,
(33:11):
I think uncanny valley definitely gets into this territory. I'm
specifically reminded of Sophia, the smiling humanoid robot. It's becomes
something of a meme and it's very impressive, like that's
the thing. At the same time it is amazing, like
this thing we be worshiped as a god in the
previous dage that it smiles too fast, smiles too fast,
and it's just it's not right, and you can't help
(33:33):
but either look away or laugh. But yeah, I mean,
I wonder if there there is an element. I don't
think it's the whole thing, but I do wonder if
it's there's sort of an undercurrent in machine humor, of
a kind of nervous laughter at you know, in machines
potentially becoming more human, more threatening to our sense of
species specialness, and potentially more threatening to our safety. Like
(33:54):
we were talking about, you know, I'm reminded of something
we mentioned fairly often on the show. Actually, ever since
our conversation with our Scott Baker about this about how
you don't need terminators or matrix agents or other evil
super intelligent villains who want to exterminate humans for AI
to represent a threat to humanity. In fact, I'm much
(34:15):
more worried about the threat from stupid AI that is
powerful before it is wise, sort of like the way
that so many examples we've discussed already just plow ahead
carrying out their functions without realizing they're going the wrong way,
or they're they're stuck, or they're not making any sense.
Uh like attached that kind of logic to practical functions
(34:36):
that have power over our lives, and there can be
terrible consequences, which also kind of comes back to children.
It's like, these are the these are the humans that
will one day rule the world, and uh and just
listen to the weird things that keep saying to us, right, Yeah, well,
I guess there's always kind of a terror at the
upcoming generations in there because there's always a little bit
of lack of understanding. Oh yeah, for sure, I mean
(34:58):
kids are always going to be amusing, but also, yeah,
you want you wonder, you know, if they're going to
be able to really grow out of this, and then likewise,
teenagers are always going to be terrifying to the older generations,
right well, So bringing it back to neural networks generating text,
one of the things that I see come up most
often in like articles about the subject is fears about
(35:19):
machine generated news articles, like specifically machine generated fake news
or hoax news articles. Of course, I'm not talking about
the version of fake news that has become a politicized term.
I mean like articles that are generated to contain false information, right,
even intentionally uh created it's a propaganda or misinformation, yeah,
(35:40):
or just nonsense. And so it shouldn't come as a
shock that they're already programs that can and do right
fake articles from AI generated text. The problem is not
that these programs exist like humans, right, hoax and propaganda
articles all the time. It's not that we couldn't create
them without machines. My fear is maybe about volume. Like,
(36:01):
if the process of putting out hoax nonsense articles and
stuff can be automated, you can just do so much
of it. And one of the keys is that it
doesn't have to be good or convincing to be damaging.
If you simply flood the zone of people's Internet experience
with obvious nonsense and trash, they don't have to believe it.
(36:24):
It can just do lots of damage to a culture
simply by flooding out everything else, right, I mean you
can you can almost imagine a future generation looking back
at a time like that and saying, well, you know,
it's really difficult to know exactly what they've they felt
in thought because of all the nonsense that was created, say,
not not just textual but video nonsense, photoshop nonsense like
(36:48):
wh where are the real people and their actual concerns
and all of this. Yeah, I really worry about the
the potential for dumb, context ignorant algorithms that seemed perfectly
capable of steering user is intellectably into more and more
horrible in mind destroying content on platforms like YouTube, you know,
and YouTube content serving algorithms that's already the world destroying AI.
(37:10):
But here's the thing, Joe, what if it's all funny?
What if it's all really funny? Well, in some cases
it is like top shelf ClickHole funny, Then then I
don't know, maybe them all in for the age of unreasoned.
I worry about this more than the kind of mind
ravaging dumb ai, way more than I worry about the
great basilisk I agree with you there. All right, Well,
(37:30):
well I want to bring it back to lighter territory here.
Uh the joke telling capabilities of Siri on an iPhone. Now,
imagine anyone out there who has an iPhone and probably
there's some degree of this in any kind of you know,
voice interface system, Alexa or or what have you. But
when you when you ask Serri to tell you a
knockdot joke, she will tell you a knockdoc joke, and
(37:51):
some of them are pretty funny. Um, and then she'll
you you ask her other things so they'll be a
hilarious response. Like if you ask her about how nine
thousand she has a funny response that's you know, preprogrammed. Again,
it's it's something that somebody wrote and created. This is
human human augmented humor. Yeah. But one of my favorites
is is one that Siri told me where clearly it
(38:12):
was there was an error that took place, and I'm
not perfectly replicating it here, but it was something like this, uh,
knock knock. I say who's there? And then Siri says Sophia,
And I say Sophia, who, and then she says, I'm sorry,
there's nobody in your call list or your contact listen Sophia.
And so I never get to find out what the
rest of that joke is because in asking, in responding
(38:36):
to the knock knock joke, she thinks I'm trying to
make a call. And so that that makes me laugh
every time that I've heard it. That's really good. It's
like a performance art kind of joke. Yeah. So it's
it's trying to be, you know, actual funny joke telling,
but then it gets into this machine error level of
joke telling, and it's probably the more hilarious outcome of
(38:57):
the two. There is no one called big Cat, big Cat,
big Cat, big Cat in your address, but there should
be that I should create that. All right, Well, there
you have it. Um, I feel like we have completely
explained humor and artificial intelligence. We hope you never find
anything funny again. Uh yeah, as so. As always, if
(39:19):
you want more content, you want more episodes of stuff
to blow your mind, head on over to Stuff to
Blow your Mind dot com. That is the mother ship.
That's where we'll find all the episodes. That's where we'll
find links out to social media accounts. That's where we'll
find a link to our t shirt store where you
can get the all Hail the Great Basilist t shirt. Uh.
So you can. You can have a little fun, have
a little uh, find a little humor in the idea
(39:40):
of future of hyper powerful AI Overlord scoring the machines.
That's right, it's a fun way to support the show.
You can also support us to the best way to
support us is just to rate and review us wherever
you have the power to do so. Like whatever platform
you're using to listen to podcasts, make sure you've thrown
a few stars in a nice review our way. Hey,
you subscribe to in Engine yet If you haven't, that's
(40:01):
our other podcast. We bring the same kind of approach.
We bring on stuff to blow your mind, but we
talk about inventions from history. If you like this show,
we're pretty sure you'll like that one too, so you
should go check it out and subscribe. Yeah, two whole
episodes on toilets, for instance. Oh man, that was fun.
Never look at a toilet the same way way again. Honestly,
I don't I have way more appreciation of toilet kind
(40:22):
than I did before. And there's an example technology that
is inherently funny and inherently good. Yeah, it's very good,
very important. Anyway, huge thanks to our excellent audio producers
Alex Williams and Tor Harrison. If you would like to
reach out to us, to send us an email and
let us know what you think about the show, feedback
on this episode, or any other suggestions for future topics,
(40:44):
or just let us know maybe how you found out
about us where you listen from that kind of stuff.
You can email us at blow the Mind at how
stuff works dot com for more on this and thousands
of other topics. Is it how stuff works dot com
Welcome to Stuff to Blow Your Mind from how Stuff
Works dot com. Hey, welcome to Stuff to Blow your Mind.
My name is Robert Lamb and I'm Joe McCormick, and
we're back with part two of our discussion of the
plate a sex Mac and I think that's what we're
gonna call it, the comedy of errors that exist in
(00:25):
the machine and artificial intelligence world. That's right. We were
gonna do it as one episode, but it just went
too long, had to break it up into two. But
it has everything it has, uh, it has humor, it
has theories of humor, it has killer robots, references to RoboCop,
everything you could ask for. One of my favorite things
explaining a joke to death, yes and then yes yeah. Yeah.
(00:49):
Basically this is a lot of joke explaining. But I
think by the end of it, everyone will still retain
their ability to laugh. All right, So, in thinking about
what's so funny about machine failure, uh, and and even
machine success along a certain you know, parameter of distance
from what from what it's ultimate level of success would be,
you know, AI, that's sort of succeeding but still isn't
(01:11):
really there. Um, I guess it's worthwhile to look at
what are some of the main theories out there about
what humor is? Like, why do we find anything funny
not just machines? Yeah, because it becomes increasingly difficult to
really uh nail it down, you know, you come back
to that whole idea of it's funny because it's funny
or you know what when you see it. Uh. This
(01:33):
topic has come up a few times on the show
before um most recently two thousands, seventeens laughing during horror movies.
That was when Christian was on the show and we
talked about you know, what is it about horror movies
like frightening, terrifying films that that elicits laughter from us?
And then there were a few other ones in the past,
The healing power of laughter, the killing joke, funny or Die.
(01:54):
These are all episodes of stuff to blow your mind.
You can you can check out in the vault if
you so desire. Yeah, and so what if you go
and check those out, I'm sure you as you'll probably
found before. There's no one theory of human right and
nobody agrees, but there's sort of a host of competing
theories that I would say aren't even always totally contradicting
(02:16):
of one another. They they sort of partially overlap and
partially contradict one another. Right, Yeah, there's not really a
theory of everything per se, but like each one feels
a little bit right, like each theory is is kind
of fondling the elephant. Yeah, exactly. Uh So, a brief
run through of a few at the top. And there's
no way to explore all the theories of humor. But
(02:37):
if you are the most commonly cited and most popular ones,
one would be the superiority theory. And this is propounded
in some classical works, like in the works of Plato
and Aristotle. This one is actually is kind of nasty,
But sometimes I guess humor is kind of nasty. It
proposes that we laugh when we notice someone is less
fortunate than us in one way or another, and the
(02:58):
laughter comes from our feeling that we occupy a place
of superiority. We sort of deploy it as a form
of scorn on somebody else. And Plato and his Philibus dialogue,
which is discussing the nature of pleasure and different kinds
of pleasure, like why pleasures of the mind might be
superior to pleasures of the flesh. Plato has Socrates claim
(03:20):
that we laugh at people who do not recognize their
own misfortune, for example, when people are stupid but think
themselves brilliant, or when people are ugly but think themselves
very handsome. And I would say this clearly doesn't cover
all types of humor, Like, you know, what about self
deprecating humor when you can laugh at yourself, But there's
(03:40):
something there that is clearly present in some humor. Yeah, well,
you know, when it comes to self deprecating humor and
laughing at yourself, I think it does make sense when
you consider our abilities to sometimes step outside of ourselves
and see ourselves as a character in a story, sort
of the infectious nature of narrative there. But then also
in the way that we use theory of mind way
(04:01):
too much to try and imagine how others see us.
So maybe, like, if you're trying to roll with this theory,
you'd say that self deprecating humor is almost when we
pretend that we are ourselves someone else and you like
step step back and you scorn a different version of yourself. Yeah,
basically like or you're just kind of imagining how scornful
other people were being to you because of something you
(04:24):
did or said or thought. Yeah. Um, I mean another
thing I would just say is that unless you're a
horrible person, and maybe you know, Plato and Aristotle were
in some ways very smart but also very horrible people,
unless you're a really bad person, you don't find most
types of misfortunes of other people funny. I mean, I
do see a certain distinction along Plato's lens in the Philippus,
(04:46):
where like a person just failing to do things that
you could succeed at is generally not very funny. Like
somebody trying to solve a math problem you could solve
and they can't do it, that's not funny. But like
if a man proclaims himself to be one of the
greatest brains of all time and then is making basic
errors of logic and math, that's usually funny. Yeah. I
(05:08):
think a lot of it hinges on just you know,
what are the stakes and what sort of misfortune is
taking place? Like Another sort of example of this is
when someone thinks they are refined about something, uh, and
maybe they're a little snobbish about it but there, but
it turns out their tastes are not quite as refined
as they thought. Um. I think there was an episode
of Three's Company back you know in the day, in
(05:31):
which a don Knots character is ordering like a fine
scotch at a bar and he's really talking it up
like he knows what he's talking about, exactly what kind
of scotch he wants, and then when the bartender finally
serves it to him, he requests that the rest of
the glass be filled up with root beer, and you know,
it's like, like, that's that's so that's hilarious because he
turns out he's not as as refined as he thought.
But then it's also like we were being snobbish in
(05:54):
making that judgment. Uh, And it's and that's always kind
of a dick move, right because in anything like whatever
your thing is, that you feel like you're refined and
be at food or drinks or film or literature, like
there was a point in your life when you were
not when you were the person that was the subject
of scorn. And you know, ultimately we're being fair. You
should celebrate that they are at all interested in the
(06:15):
in this thing at all and that they are on
hopefully on some sort of the same journey that you are,
but instead we often laugh and said, can you believe
that guy just put root beer all in that sky?
I think that's a really good point. So I don't know.
I would say I definitely see something going on here
that there are definitely elements of this in humor, but
(06:37):
I don't think it explains all humor. I don't know
if maybe, maybe, maybe not. I don't know if it's
a good fit for why machines are funny, because obviously
machines aren't people exactly, but maybe we sort of see
them as people when we think they're funny. Yeah, I
think if we if you factor in the actual and
or perceived abilities of AI and robots, either either what
(06:58):
they can actually do or more I think, to the
point what our science fiction and our futurist scenarios have
proposed that they will do or could do one day,
because we end up with this situation where like we
watch our room but do something stupid, and there's it's
hilarious because like, ha, you're never going to terminate or me,
like you can't even you can't even deal with the
(07:20):
uh with this rug or this pile of cat vomit. Well, actually,
though I would say another reason this does kind of
connect is it does kind of connect to Plato's thing
about laughing atself ignorance, because one of the funniest things
that I sort of identified earlier about like the D
two O nine scene and all that, one of the
funniest things about the ways that machines fail is the
(07:41):
way that they are just so completely oblivious to their
own shortcomings, the way they just uh, the way I
think of it is that they fail and then plow
straight ahead. You know, they have they have done something
crucially wrong, and they do not appreciate this fact. All right,
let's keep rolling. What are some other theories can discuss? Okay,
next one would be like the nervous energy release theory.
(08:04):
And this is not surprisingly at all a take that
was popular with Freudians. Under this model, laughter is sort
of a pleasurable release of built up psychic tension caused
by fear or apprehension. Uh. And there's a certain kind
of logic to this, right in the way that's a
good Comedians repeatedly build at tension and then dissipated in
various ways. I bet this also goes with the laughing
(08:26):
during horror movies thing, you know, Yeah, for sure, Like you,
you get your tension built up when there's something scary
is about to happen, and then after it's over you
you release the tension and you can laugh. Yeah. Certainly
with with comedians who really leaned into like the awkwardness
and the tension, like say like an Andy Kaufman. You
know where their whole their whole bits or if you
(08:46):
want to call them bits, where it's really just making
everyone uncomfortable. It's not even about punchlines. It's about just
creating this this tension that at some point has to
be relief. Like at some point all you can do
is laugh. Yeah, So there's something there. I I this
clearly doesn't account for everything that's funny, you know, it
doesn't account for puns and all that kind of thing.
Uh It, I don't know if I can see a
(09:07):
way that it really fits with why technology failing is
all that funny? No, Yeah, I don't think. I don't
think there's anything particularly insightful about this theory regarding what
we're talking about here today. Okay, here's another one. How
about the adaptation, uh to signal play so that laughter
would be an evolutionary adaptation that occurs when we're engaged
in pretend aggression during play behaviors. It occurs as a
(09:30):
signal to help everybody involved understand the fact the behaviors
like chasing and wrestling and stuff are meant as play
rather than as genuine aggression or harm. Yeah, this one
I think is important to mention here, if only for
the reason that it touches on the social nature of laughter,
that that laughter is also about communicating with other people
(09:51):
regarding various threats, etcetera. Um. It's also I wonder if
this is one where our storytelling has kind of kind
of led us astray as well, though, because how many
um individuals that actually intends harm are laughing hysterically or sardonically.
Like I'm sure it happens, I mean, and maybe one
(10:14):
or two scenarios even come to mind where someone was
did something horrible and violent and it was seen laughing,
But I don't feel like it probably occurs anywhere as
near as often as it does in movies where villains
are constantly laughing. Be it, you know, an extreme example
like the Joker, or or just any you know, you know,
sardonic villain who just doing maniacally laughs while attempting to
(10:37):
destroy the world. Uh yeah, I mean I would say
this probably does go along with the theory then. I mean,
if it if somebody is laughing while they're actually attempting harm,
there you know their victim is not laughing so like,
so yeah, I think this goes along there. Like and
if they are, then it's like that scene in the
original um what was a Little Shop of Horrors where
(10:58):
or maybe it was in the remake is well too,
where the uh, the the the the the evil dentist
is going to work on this person and this person
just wants it, They want the pain, and it just
totally messes up his his vibe. Yeah, I haven't thought
how it would accommodate the idea of like villains laughing
while they do what they do. But I mean, I
think this is clearly true in some cases that people
(11:18):
laugh during mock aggressive play behaviors. That dogs wag their
tails during mock aggressive play behaviors. It's like, you know,
they're growling and snarling and tumbling, but you can tell
they're not really fighting because their tails are wagging. And
this is a nice lead into one of one of
my favorite theories, and this is one that we discussed
on the show before in the past, and that's the
(11:39):
benign violation theory. Yeah, and this theory is very popular now.
This theory proposes that humor occurs when one we perceive
a way in which rules or norms of some kind
are broken. And these rules can be any, you know,
all kinds of things. They can be grammatical rules, they
could be biological classifications, they can be the rules of
(11:59):
making sense, they can be moral norms. Uh. And then second,
we recognize that there has been a violation, but it's
not harmful, it's benign in nature. And then third we
see this contradiction between the fact that there's a violation
and the fact that it's harmless, and that contradiction triggers humor.
Like we said, this one is very popular. It's supported
(12:20):
by some empirical research. Just one example of research and
support that I came across as a two thousand ten
study in Psychological Science by McGraw and Warren called benign
violations making im moral behavior funny and too so. To
just cite one example that they talk about within that study, quote,
people who are more weakly committed to a norm can
(12:41):
recognize the violation, but are less likely to be threatened
or to directly experience the violation's repercussions. Consider a news
story about a church that raffles off a hummer suv
as part of a promotion for its members. Engaging in
such a secular promotion jeopardizes the sanctity of the church.
End Although most people consider churches sacred, churchgoers should be
(13:04):
more strongly committed to this belief than are people who
do not attend church. So the researchers looked at this
and they found that while churchgoers and non churchgoers alike
were about equally disgusted with this story, you know, the
idea that a church would be trying to get people
to join by giving them a chance to win a car,
non churchgoers more often found the story humorous, and so
(13:27):
that was like non churchgoers to churchgoers finding it humorous
was nine percent to sixty two percent. So the idea
is that non churchgoers saw this as a benign violation
and thus funny, and churchgoers saw it as a real
harmful violation and thus not funny. Yeah, and I think
that's something that's important to keep in mind. And you know,
(13:47):
when when looking at the benign violation theory is that
how ultimately subjective it is going to be. And of
course that is comedy as well, so it matches up
in that regard. Right, It depends on whether you actually
see something as a violation and whether you actually see
something as harmful or not. Now, there are different ways
you could play around with the definitions and of this
theory to like, like, you could make it very elastic
(14:10):
to accommodate a lot of stuff, or you could more
narrowly tailor it. I would say one problem potentially with
it is that lots of things that are benign violations
aren't funny at all. Uh. An example I came up
with is, imagine you are not supposed to take home
office supplies, but you do you take home a pin. Well,
that's a violation of the rules. It doesn't it's not
(14:32):
really harmful, it's benign, and it's not really funny. Right.
But on the other hand, I would I think the
counter argument here would be, okay, taking one pin home
from work, like that's I do that all the time,
most of us do. It's one thing to take a
pin home, even accidental, accidentally, take one home to you know,
purposely take one home. But what have we upped it?
What have instead of accidentally taking a pin home from work,
(14:55):
you accidentally took all the pins home from work. That's
funny now that it was getting into you know a
little more. And I'm not saying that there's it's the
funniest gag of all time, but certainly there's more room
for hilarity in this scenario. Well, that starts to be
more like a machine type comedic mistake because it's so uh,
it's so over the top. It seems to kind of
(15:16):
like not just violate a norm but just plows straight
through it and goes all the way down right where.
I guess another way that would work is if one
meant to steal something of greater value, like one intended
to to um to perpetrate some greater violation of office norms,
and then in the end you only stole a pin,
(15:36):
but even in a done land as well, and it
really needs to be a box of pins or all
the pins. Yeah, I guess you could say. I mean,
I think you could probably play around with exactly how
you define the terms of benign violation to maybe accommodate
those counter examples. I would say also though lots of
things that are funny too, people can feel like they're
not at all benign, and like think of videos of
people falling off skateboards and smashing their faces or the
(15:59):
crotches and the handrails and brick walls and stuff. People
find these things quite funny, and people can get really
hurt in them. Yeah, I think with with that, I
think part of it is that we generally see confirmation
that they survived, or if we don't see confirmation that
they survived whatever, they're horrible, you know, skateboarding accident was,
then we at least don't see them going to the
(16:20):
e er, you know, like we have no additional information
to go on, no additional context, and we can just
kind of assume that they were okay. So, and I
also was wondering about this. I wondered what extent comedic
interpretations of slapstick injuries in which harm generally doesn't extend
beyond the punch line. Um, if these program us to
(16:41):
to make these kind of comedic interpretations. Uh, for instance,
you know, when you think about comedic violence, you think
about the Three Stooges, and certainly they're just always brutalizing
each other and then they're fine, and then they're back
for another adventure. They're fine. You don't worry about their health.
Take a good old fashioned three Stooges fingerpoke. Um, you
(17:02):
know we have in the in the eye do you
have two fingers and then you go right for the
eyes and if nobody gets the block up, bam, there's
like as more of a blank noise, right. So yeah,
so when the Stooges do it, uh, it's hilarious, right,
there's a lot of slapstick hilarity with with the double fingerpoke.
But and then likewise you look at something like pro wrestling,
(17:24):
in pro wrestling, which of course is this like sports
theater thing. Uh, the fingerpoke is a comedy spot, you know,
it's it's if someone goes for the double eye poke,
it's going to be a moment of hilarity as well.
But if somebody does it in a basketball game, that's
not really funny, right, or or if someone does it
in most if not all combat sports, it is it
(17:47):
is a band maneuver. Like generally, if you're going for
somebody's eyes, like that's either it's either really evil or
it's a move of desperation. Yeah, this does certainly make
that to whatever extent the benign violation theory is true,
it's highly context dependent and and maybe the truth there,
if there is something to this theory is that um
(18:07):
is that the context makes us determine what's benign and
what's not right, and it's gonna be highly dependent on
the culture you're in, the time you live in. I mean, certainly,
just thinking of workplace humor, there are a lot of
violations today that are serious that may have been perceived
as more benign in days past, or as or seen
(18:27):
as not violations at all. Oh yeah, this go back
to like the nineteen eighties, and like workplace sexual harassment
was a common like sitcom gag at the time. It's
just like, oh, it's hilarious, you know, he's he's harassing
Tina again. Uh, And now like that's not really funny
at all anymore because we see that, I think, I
think because people see that as a more genuine violation,
(18:49):
not as something that's just a harmless you can wave
it off, that's just ted being ted. So how do
you feel about about benign violation and neural networks? Well, strangely,
I do see part of a fit here. Um, because
I along the lines of the Church suv example. The
more serious and reverent the context in which the technology
(19:10):
is presented, the funnier it is when it fails. I
think about like ED two oh nine. This the scene
is sort of made by the fact that Dick Jones
and all the biz bros Are proudly boasting about this
marvelous new technology that will be the future, and then
it fails in such a spectacular way. I think in general,
the failures of AI are much funnier when they're presented
(19:32):
within a context of us knowing about people who believe
in AI messiahs and you know AI Basilisks and Satan's
and all that who who uh see this great future
of of power and dominance among Aiyes, watching them mess
up in hilarious ways becomes all the more poignant and perfect.
All right, time to take a quick break. Well, we
(19:54):
will be right back. Thank alright, we're back. So one
more theory is the in congruity resolution theory. This is
a theory that's had various forms over the years, for
example advocated by Manuel Kant and others, but essentially the
idea here is that there is an incongruity between expectations
and reality. Laughter occurs when we realize this incongruity can
(20:17):
be resolved, and this is there's clearly a strong element
of this in lots of humors. Again, what joke involves
a set up and then a punch line, and most
often the set up sets you up to expect one
type of thing. The punch line subverts your expectations. So
that's clearly there. But then again, there are lots of
things that are surprising and turn our expectations on their heads,
(20:39):
and they're not funny. Yeah, like we incongruity has to
be novel. Yeah, you know, it can't be a situation
where I thought I was buying mozzarella, but I really
got monster. That's not that. That's really not funny. Now,
it would be funny if you thought you were buying
mozzarella you got home and it was like rat eggs,
especially because rats don't lay eggs. Yeah, it's like that. Yeah,
(21:00):
that would be that would be a lot more of hilarious.
One different but somewhat similar theory. I was reading about that.
I'd like to read this book now that I know
it exists. It's a two thousand eleven book from M
I t press called Inside Jokes by the author's Matthew Hurley,
Daniel Dennett, and Reginald Adams Reginald Adams Jr. And they
(21:21):
argue against the sufficiency of any of these models we've
talked about already. They propose instead that humor is an
evolutionary reward system. It's a pleasurable reward feeling that we
get when we recognize the inappropriateness of a mental representation
and then fix it. So it's the brains. It's it's
an internal reward system like we get from other things
(21:43):
that are good for the body, like eating or something. Uh.
It's the brain's reward system for debugging itself. And so
you can imagine this is why many jokes have a
set up. The setup prepares you to establish an inappropriate
or incongruous mental representation. The punch line suddenly creates that
inappropriate representation, and then the debugging that immediately happens in
(22:06):
your brain following is rewarded with this pleasurable feeling of humor. Interesting.
I think that's something that that probably matches as a
model that matches with a lot of our experience of
of having hilarious or ridiculous notions enter our head and
then you just miss them, or you even don't, you're
not even considering them, but they just rise up from
(22:27):
the depths of your psyche. Yeah, and I think it
kind of fits well with what's funny about each entry
in these like machine generated pieces of text, because each
moment you're going through this text, you're sort of being
set up to think about the text in a certain
way given the category you're already looking for, like names
of spells and stuff. Then constantly you're given something that
(22:49):
doesn't really fit, and then your brain has to debug
why it doesn't really fit, and that feels good. I'm
not sure it's a perfect theory that covers all of
humor either, but I am interested in Maybe you know,
another thing is that, like you said earlier, it's possible
there's no single, perfect, overarching theory of humor, Like what
if humor is a category held together not by a
(23:10):
single all inclusive set of criteria, but by a Wittgensteinian
kind of family resemblances principle. There's no one thing that
that all humor is. Humor is instead multiple different related
sets of things, none of which are entirely contained in
a single box. By the way, you know, if you
open that box, it's going to be one of those
(23:30):
spring snakes that comes flying out, Like open the fake
cannon nuts. Which one does that fit in? Um? I
guess you know. It has to do with expectation, but
also it's benign violation. So if you're expecting nuts, not snakes,
but you've got a snake. But then, well it's not
a real snake, which is a perfect animal to include
in that gag given our you know, our our evolutionary
(23:51):
response to a serpentine form. What if it's a real
snake that bites you on the throat, I would like
to see that happen more or what. One of my
favorite gags that I did this is that a workplace
at some point there was this I had to like
kind of this lame boss who had a can. He
had this exact gag of fake nuts, fake peanuts, and
if you opened it, a snake jumped out. Um. So
(24:14):
one day he was gone for the day and so
I took the snake, got rid of the snake, and
I filled it with actual peanuts, and it gave me
so much pleasure. I don't I never even got to
witness the payoff. But I just love the idea of
him going to like pull this out on somebody or
to just amuse himself, and then it's actually filled with nuts.
And I don't know how lame the boss was. This
was This is kind of a time in my life
(24:35):
when all bosses were lame. So if any of my
previous bosses are listening to me, I'm probably not talking
about you. Well, there's another theory of humor that kind
of fits that because of the way in which humor
sort of interlocks with our expectations of people based on
their individual characteristics. Uh So, the next one clearly doesn't
account for everything that's funny, but it does have a
(24:56):
kind of elegant, elegant connection to what we're talking about.
So this is a theory of humor put forward by
a philosopher named Henri Bergson in a book called Laughter,
published in nineteen hundred and One of the ideas Bergson
puts forward here is that laughter is often our response
to rigidity, inflexibility, and failure to adapt. And this is
(25:17):
one of the reasons why comedies involve us being aware
of like the particular flaws of a character and then
seeing those same flaws acted out over and over again.
Think about sitcoms, like, it's not just that the characters
have failures, but they tend to fail repeatedly in in
individually characteristic ways. Homer Simpson is lazy, and he's thoughtless,
(25:40):
and he's motivated by donuts and gluttony and all that. Uh.
Michael Scott in the office is needy and attention seeking
and unaware of how he's perceived by others. Characters have
these sort of trademarks and they keep making the same
kinds of mistakes over and over, and it's their inability
to adapt that makes humorous in this point of view. Now,
(26:01):
clearly this doesn't account for everything that's funny, but it
sort of ties in loosely with the Hurley Dinnett Adams
theory that humor rewards adaptation and mental debugging. These people,
if they're making the same mistakes over and over, they're
not debugging, and you're like debugging by watching them. But
then again, of course we have to come back to
sort of the benign violation theory a little bit here,
because we don't want characters that uh failed too much,
(26:25):
you know, like you don't want it to be like
a tragic level of failure, right yeah. And of course
in reality people fail to adapt all the all the time,
and ways that aren't funny but are actually like sad
or tragic or you know, it's something very bad about it.
And then there's that whole sliding scale too, because especially
with something like like Michael Scott, especially like the Office
got into so much awkward territory just every week, week
(26:47):
after week, and um, a lot of that really got
into sort of painful, pitiful territory, so that they definitely
changed exactly where the line was in the sand, but
but still ultimately, like you know, Michael Scott didn't die
at the end of an episode, well, it helped by
coming back and making the character lovable, like he wasn't
(27:09):
getting into that like sad awkwardness and then just ending
in a place where you hate him, right yeah. But anyway,
I think this theory about like rigidity and inflexibility and
like repeating the same types of mistakes without adapting has
some strong purchase on the funniest features of machine failure comedy.
It's not just that the machine fails, but that the
(27:31):
machine fails repeatedly mechanically. According to inflexible guidelines, like the
way it just keeps following the rules, when a neural
net text generator spits out not just one nonsensical phrase,
but like reams and reams of text without acknowledging or
knowing it all that the first few words didn't make
any sense. Or when I think about when a glitching
(27:52):
video game character is stuck inside a table and just
keeps running in place and talking. That's one of my favorites.
Or that lack of contacts, lack of context, lack of
self awareness, and lack of adaptability. The way machines just
plow ahead despite problems that are obvious to us that
they're clearly not aware of. I think that's key to
(28:14):
what makes them so especially funny. And in this sense,
they're kind of like the Homer Simpsons and the Michael Scott's,
the characters with inflexible, repetitive behaviors that lead to disaster
but which they never learned from or change. You know,
I can't help but come back to the idea of
of a child and all of this and the child
analogy here, so you know we can we can laugh
(28:34):
at our own children's mistakes are weird interpretations because we
know that they're they're largely going to adapt, they're going
to grow up, uh, and that they're going to leave
behind their their dragons or hopefully they're just gonna remember
how to summon them in appropriate ways now. But if
they're the robot is hilarious because it continues to fail,
and of course also because it is a machine and
(28:56):
not a human child. Perhaps part of the issue too,
is that a machine is but a prototype, and it's
next phase of advancement does not necessarily exist within itself,
but in the next prototype and another individual, if you will.
I think that's a really good point. Yes, Like, so
a child can make a funny mistake, but then the
(29:16):
child immediately within themselves learn something and or you know,
at least as the potential learn something and grow and
get better at not making that kind of mistake anymore,
which in some cases kind of tragic because the mistakes
are glorious. Oh yeah, like half the in jokes that
my wife and I make are two things that my
my son has not said in years, but he said
(29:38):
it was and it was hilarious, and now we repeat
it to each other. But like most of these machines,
even though they might have the ability to learn from
their mistakes. They learn from their their mistakes in like
new instantiations of one of themselves, like you're saying, it's
the next prototype. They don't automatically learn from their mistakes.
You have to do something to them to reboot them
(29:59):
or retrain the them or something right. And if it's
a video game that, yeah, this, this character is gonna
keep getting stuck in that table. Maybe there'll be a
patch that fixes it, but for the most part, it's
going to be the sequel that tries to get it right.
All right, we're gonna take a quick break, but we'll
be right back. And we're back now. I think this
is all one of one of the reasons that the
(30:19):
Black Mirror episode Metal Head from is so effective. Oh
it's it's it's one of my favorites. Um this was
This was again seen. And this came after years of
us watching impressive, yet ultimately clumsy and imperfect Boston Dynamics
walking robot videos, videos of such you know entities as
(30:42):
Big Dog, you know, which kind of has this springy
brant kind of a sound as it prances along, and
it's very impressive to look at and yet at the
same time, you know their videos of being kicked around,
and you know, certainly if it's succeeding a lot, but
also not quite it's not reaching that level of like
mature technology. Yeah, though I worry how much I already
(31:05):
sympathize with it, Like watching those Boston Dynamics robots when
somebody kicks the dog. I know it's just a robot,
but I don't like that person who kicked it. I'm like,
that's cruel. How could you do that? Well? In the
Metal Head, Uh, Basically, the whole thing is kind of
a post um apocalyptic scenario, killer robot scenario, and the
(31:25):
killer robot we see is not a juvenile prototype, but
what you might classify as an adult that the technology
in a proceed at a perceived level of maturity. So
the Boston Dynamics Big Dog all grown up into a
terrifying lethal, quadrupedal killing machine that it overcomes every challenge
(31:48):
in the natural or man made world that it is
presented with. Uh. And I think that's that's one of
the reasons it is terrifying, because it's kind of like, oh, yeah,
these things are going to grow up. We're for a
lot of money and effort into making them grow up,
and uh and and how are we going to relate
to that technology when it reaches that point? Well, that
brings us back to something you were saying earlier about
(32:09):
like the CS Lewis quote, Like clearly our machines and
their failures I think are becoming much funnier over time.
Like a steam engine designed to pump water out of
a coal mine is really not funny when it malfunctions,
maybe like cut somebody's arm off, like it's it's not hilarious.
Automatic doors are still not very funny. Maybe a little
(32:30):
bit funnier, like I was saying, when they're trying to
close on something over and over, and you know when
it gets repetitive like that room buzz are funnier. Anything
attempting to generate real human language or simulate direct human behaviors,
like say the DARPA challenge bipedal robots. You know when
you watch them like collapsing in the sand, or like
(32:50):
when one of them like opens a door and the
door goes open and then they just crumple in a
heap on the ground and that kind of thing. It
is definitely even funny to hear, like machines seem to
get funnier. The closer they get too resembling real human behaviors,
the better they simulate. And I wonder, so it's almost
like there's an uncanny Valley of humor in a way. Yeah,
(33:11):
I think uncanny valley definitely gets into this territory. I'm
specifically reminded of Sophia, the smiling humanoid robot. It's becomes
something of a meme and it's very impressive, like that's
the thing. At the same time it is amazing, like
this thing we be worshiped as a god in the
previous dage that it smiles too fast, smiles too fast,
and it's just it's not right, and you can't help
(33:33):
but either look away or laugh. But yeah, I mean,
I wonder if there there is an element. I don't
think it's the whole thing, but I do wonder if
it's there's sort of an undercurrent in machine humor, of
a kind of nervous laughter at you know, in machines
potentially becoming more human, more threatening to our sense of
species specialness, and potentially more threatening to our safety. Like
(33:54):
we were talking about, you know, I'm reminded of something
we mentioned fairly often on the show. Actually, ever since
our conversation with our Scott Baker about this about how
you don't need terminators or matrix agents or other evil
super intelligent villains who want to exterminate humans for AI
to represent a threat to humanity. In fact, I'm much
(34:15):
more worried about the threat from stupid AI that is
powerful before it is wise, sort of like the way
that so many examples we've discussed already just plow ahead
carrying out their functions without realizing they're going the wrong way,
or they're they're stuck, or they're not making any sense.
Uh like attached that kind of logic to practical functions
(34:36):
that have power over our lives, and there can be
terrible consequences, which also kind of comes back to children.
It's like, these are the these are the humans that
will one day rule the world, and uh and just
listen to the weird things that keep saying to us, right, Yeah, well,
I guess there's always kind of a terror at the
upcoming generations in there because there's always a little bit
of lack of understanding. Oh yeah, for sure, I mean
(34:58):
kids are always going to be amusing, but also, yeah,
you want you wonder, you know, if they're going to
be able to really grow out of this, and then likewise,
teenagers are always going to be terrifying to the older generations,
right well, So bringing it back to neural networks generating text,
one of the things that I see come up most
often in like articles about the subject is fears about
(35:19):
machine generated news articles, like specifically machine generated fake news
or hoax news articles. Of course, I'm not talking about
the version of fake news that has become a politicized term.
I mean like articles that are generated to contain false information, right,
even intentionally uh created it's a propaganda or misinformation, yeah,
(35:40):
or just nonsense. And so it shouldn't come as a
shock that they're already programs that can and do right
fake articles from AI generated text. The problem is not
that these programs exist like humans, right, hoax and propaganda
articles all the time. It's not that we couldn't create
them without machines. My fear is maybe about volume. Like,
(36:01):
if the process of putting out hoax nonsense articles and
stuff can be automated, you can just do so much
of it. And one of the keys is that it
doesn't have to be good or convincing to be damaging.
If you simply flood the zone of people's Internet experience
with obvious nonsense and trash, they don't have to believe it.
(36:24):
It can just do lots of damage to a culture
simply by flooding out everything else, right, I mean you
can you can almost imagine a future generation looking back
at a time like that and saying, well, you know,
it's really difficult to know exactly what they've they felt
in thought because of all the nonsense that was created, say,
not not just textual but video nonsense, photoshop nonsense like
(36:48):
wh where are the real people and their actual concerns
and all of this. Yeah, I really worry about the
the potential for dumb, context ignorant algorithms that seemed perfectly
capable of steering user is intellectably into more and more
horrible in mind destroying content on platforms like YouTube, you know,
and YouTube content serving algorithms that's already the world destroying AI.
(37:10):
But here's the thing, Joe, what if it's all funny?
What if it's all really funny? Well, in some cases
it is like top shelf ClickHole funny, Then then I
don't know, maybe them all in for the age of unreasoned.
I worry about this more than the kind of mind
ravaging dumb ai, way more than I worry about the
great basilisk I agree with you there. All right, Well,
(37:30):
well I want to bring it back to lighter territory here.
Uh the joke telling capabilities of Siri on an iPhone. Now,
imagine anyone out there who has an iPhone and probably
there's some degree of this in any kind of you know,
voice interface system, Alexa or or what have you. But
when you when you ask Serri to tell you a
knockdot joke, she will tell you a knockdoc joke, and
(37:51):
some of them are pretty funny. Um, and then she'll
you you ask her other things so they'll be a
hilarious response. Like if you ask her about how nine
thousand she has a funny response that's you know, preprogrammed. Again,
it's it's something that somebody wrote and created. This is
human human augmented humor. Yeah. But one of my favorites
is is one that Siri told me where clearly it
(38:12):
was there was an error that took place, and I'm
not perfectly replicating it here, but it was something like this, uh,
knock knock. I say who's there? And then Siri says Sophia,
And I say Sophia, who, and then she says, I'm sorry,
there's nobody in your call list or your contact listen Sophia.
And so I never get to find out what the
rest of that joke is because in asking, in responding
(38:36):
to the knock knock joke, she thinks I'm trying to
make a call. And so that that makes me laugh
every time that I've heard it. That's really good. It's
like a performance art kind of joke. Yeah. So it's
it's trying to be, you know, actual funny joke telling,
but then it gets into this machine error level of
joke telling, and it's probably the more hilarious outcome of
(38:57):
the two. There is no one called big Cat, big Cat,
big Cat, big Cat in your address, but there should
be that I should create that. All right, Well, there
you have it. Um, I feel like we have completely
explained humor and artificial intelligence. We hope you never find
anything funny again. Uh yeah, as so. As always, if
(39:19):
you want more content, you want more episodes of stuff
to blow your mind, head on over to Stuff to
Blow your Mind dot com. That is the mother ship.
That's where we'll find all the episodes. That's where we'll
find links out to social media accounts. That's where we'll
find a link to our t shirt store where you
can get the all Hail the Great Basilist t shirt. Uh.
So you can. You can have a little fun, have
a little uh, find a little humor in the idea
(39:40):
of future of hyper powerful AI Overlord scoring the machines.
That's right, it's a fun way to support the show.
You can also support us to the best way to
support us is just to rate and review us wherever
you have the power to do so. Like whatever platform
you're using to listen to podcasts, make sure you've thrown
a few stars in a nice review our way. Hey,
you subscribe to in Engine yet If you haven't, that's
(40:01):
our other podcast. We bring the same kind of approach.
We bring on stuff to blow your mind, but we
talk about inventions from history. If you like this show,
we're pretty sure you'll like that one too, so you
should go check it out and subscribe. Yeah, two whole
episodes on toilets, for instance. Oh man, that was fun.
Never look at a toilet the same way way again. Honestly,
I don't I have way more appreciation of toilet kind
(40:22):
than I did before. And there's an example technology that
is inherently funny and inherently good. Yeah, it's very good,
very important. Anyway, huge thanks to our excellent audio producers
Alex Williams and Tor Harrison. If you would like to
reach out to us, to send us an email and
let us know what you think about the show, feedback
on this episode, or any other suggestions for future topics,
(40:44):
or just let us know maybe how you found out
about us where you listen from that kind of stuff.
You can email us at blow the Mind at how
stuff works dot com for more on this and thousands
of other topics. Is it how stuff works dot com
Stuff To Blow Your Mind News
Hosts And Creators
Robert Lamb
Joe McCormick
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.