This Episode Was Written By AI (Sort Of)

Can generative AI write a good episode for a technology podcast? We find out with this experiment about airbags written by ChatGPT. What worked, what didn't work, and what does Jonathan find really troubling? 

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:04):
Welcome to tech Stuff, a production from iHeartRadio. Hey there,
and welcome to tech Stuff. I'm your host, Jonathan Strickland.
I'm an executive producer with iHeart Podcasts and how the
tech are you? So what you're about to hear is
an experiment. I decided to see what chat GPT could

(00:27):
scramble up if I asked it to create a tech
podcast episode. Now, I didn't specifically say a tech stuff
episode in particular. Instead, my prompt literally read quote, create
a tech podcast episode about airbags, end quote, and I
just chose airbags at random. This is what it came

(00:50):
up with. So, without further ado, listeners, I bring you
the following episode from chat GPT, and this will all
be quoted directly from chat GPT. You'll know when I
come back in. So here we go. Tech Innovations, air
Bags and Automotive Safety. Host Welcome to Tech Innovations, the

(01:13):
podcast where we explore the latest and greatest in technology
and innovation. I'm your host, your name, and today we're
diving into a crucial aspect of automotive safety. Airbags. These
life saving devices have been a staple of vehicles for decades,
but their technology and impact are often overlooked. Today, we'll
uncover the history, technology, and future of airbags. Let's get started.

(01:38):
Segment one, The history of airbags. HOST. Air bags have
a fascinating history that dates back to the nineteen fifties.
The idea of a cushion to protect drivers in a
crash was first conceived by John W. Hetrick, who filed
the initial patent in nineteen fifty one. However, it wasn't
until the nineteen seventies the airbags started to become a
practical reality. Guest expert joining us today is doctor Emily

(02:01):
Roberts and automotive safety historian Doctor Roberts. Can you tell
us about the early development of air bags in their
initial reception? Dtor Roberts. Absolutely, The early prototypes were quite
rudimentary compared to today's standards. They relied on compressed air
and were often bulky and unreliable. However, as the technology evolved,

(02:22):
especially with the advent of solid state crash sensors in
the nineteen eighties, airbags became more efficient and began to
gain acceptance among manufacturers and consumers. Segment two. How airbags work. HOST.
Modern airbags are marvels of engineering at their core. They're
designed to deploy within milliseconds of a collision to cushion

(02:43):
and protect the occupants. But what exactly happens during those
critical moments? Guest expert to explain the technical workings, we
have James Lee, and automotive engineer specializing in safety systems. James,
can you break down the process for us? James Lee?
Or When a crash sensor detects a collision, it sends

(03:03):
an electric signal to the airbag control unit. This unit
triggers an igniter that rapidly inflates the air bag using
a chemical reaction, typically sodium azide. The airbag then deploys
at speeds of up to two hundred miles per hour,
providing a cushion to reduce the impact on the occupants.
Segment three types of airbags host airbags have evolved beyond

(03:25):
just standard front airbags. There are now side airbags, curtain
air bags, knee air bags, and more. Each type serves
a specific purpose in protecting different parts of the body
during various kinds of collisions. Guest Expert, we're joined by
Lisa Fernandez and expert and vehicle safety systems. Lisa, could
you tell us about the different types of air bags

(03:45):
in their roles Lisa Fernandez. Of course, front air bags
are designed to protect the head and chest in a
frontal crash. Side airbags help protect the torso and pelvis
and side impacts, while curtain air bags deploy from the
roof to shield the head in rollovers. Knee airbags are
intended to prevent leg injuries. Together they create a comprehensive

(04:05):
safety net. Segment four. The impact of airbags on safety
statistics HOST. The introduction and widespread adoption of airbags have
had a profound impact on vehicle safety. According to the
National Highway Traffic Safety Administration or NHTSA, airbags have saved
tens of thousands of lives since their introduction. Guest expert,

(04:27):
we have doctor Michael Harris, a statistician with the NHTSA,
to provide insights into the data. Doctor Harris, what do
the numbers tell us about the effectiveness of airbags, Doctor Harris?
The statistics are compelling. Front airbags have been credited with
reducing driver fatalities by about twenty nine percent and front

(04:47):
passenger fatalities by thirty two percent in frontal crashes. Side
airbags with head protection have shown even greater potential, reducing
deaths in side impacts by about thirty seven percent. Future
of airbag technology. Host. As we look to the future,
airbag technology continues to advance. Innovations such as pedestrian air

(05:09):
bags and smart air bags that adjust deployment based on
the occupant's size and position are now on the horizon.
Guest expert. To discuss these future trends, we have doctor
Sarah Kim, a researcher in automotive technology. Doctor Kim, what
exciting developments can we expect in air bag technology? Doctor Kim.
One of the most promising areas is the integration of

(05:31):
advanced sensors and artificial intelligence. These systems can assess the
severity of a crash and adjust airbag deployment accordingly. Additionally,
external air bags that deploy to protect pedestrians and cyclists
are being tested, which could significantly enhance overall road safety.
Closing remarks, Host, it's clear that airbags have revolutionized automotive

(05:52):
safety and continue to evolve from their humble beginnings to
the sophisticated systems we have today. Air Bags play a
crucial role in protecting lives. Thank you to our experts
for sharing their insights, and thank you to our listeners
for tuning in, Stay safe and stay curious. Host. This
concludes today's episode of tech Innovations. If you enjoyed this episode,

(06:13):
please subscribe and leave a review. If you have any
questions or topics you'd like us to cover in future episodes,
feel free to reach us on social media or via
our website. Until next time, I'm your name, and this
is tech Innovations. Okay, there we go. That's it. That
is the chat GPT generated episode about airbags. Obviously very

(06:36):
light episode, right, not like a tech stuff episode, doesn't
last forty seven minutes. In fact, it lasted less than
seven minutes with me reading through that, although I imagine
that the different segments would probably have things like musical
transitions and such if this were a fully produced episode,
and probably each of those segments was meant to be
fleshed out right like that was not meant to be

(06:58):
a full episode. It was more like the outline of
an episode. And after I read this, I wanted to
know what sources the AI actually used in order to
craft this episode. So this is what it had to
say when I asked for that information. The episode script
was created based on general knowledge about airbags and their history, technology,

(07:21):
and impact on automotive safety. The information is drawn from
widely known facts and common sources in the automotive safety field,
such as National Highway Traffic Safety Administration in HTSA reports
and statistics on airbag effectiveness and safety data, automotive safety
engineering principles, including how airbags function and their deployment mechanisms,

(07:43):
historical records and patents such as the original airbag patent
by John W. Hetrick. More on that in a minute,
Current trends in automotive technology, including advancements in smart air
bags and pedestrian protection systems. For more detailed or specific information,
here are a few potential sources that could provide the
data used in the episode. NHTSA website and their publications

(08:07):
on vehicle safety features, automotive engineering textbooks and technical papers
on airbag technology, interviews and articles from industry experts and
researchers in automotive safety, automotive history books that cover the
development and evolution of vehicle safety features. If you need
specific references or more detailed information from these sources, I
can look them up for you. Okay, I didn't bother

(08:29):
to go into deeper detail here. Maybe I should have,
but already this is a little puzzling, right because it
says here are a few potential sources that could provide
the data used in this episode, which is pretty darn ambiguous.
You know, the sources could do that, but there's no
guarantee they will, or that they actually served as a
crucial source for the data that was ultimately pulled for

(08:53):
this AI generated episode. The plot thickens now. I would say,
on the surface, this rudimentary tech podcast episode is more
or less fine, but let's dive down a bit and
double check the homework assignment, because there are problems here,

(09:13):
some of which are factual in nature and a lot
of which I would argue are more ethical. So the
AI generated episode is correct that John W. Hettrick frequently
gets the credit for inventing the airbag. The Institute of Physics,
which in turn cites sources such as D. Sherman's article
The Rough Road to Airbags, gives a slightly different date. However,

(09:36):
the AI had said that Hettrick filed for this patent
back in nineteen fifty one, whereas the Institute of Physics
says he actually got the idea in nineteen fifty two,
after he got into an automobile accident with his wife
and child in the car. So why would he patent
something in nineteen fifty one if he didn't even get
the idea for it until nineteen fifty two. Well, this

(09:58):
one is easy to solve. All we have to do
is pop over to a patent database and search for
Hendrick's name in the inventor field and see what it
has on there. When I did that, I saw that
the date for the initial filing of a patent from
John Hendrick about airbags was on August fifth, nineteen fifty two. Okay,

(10:22):
that's one data point that's wrong, but it's only wrong
by a year, right, Like, this isn't the worst mistake
that AI could make. It's the right person, it's the
right invention, it's just the wrong year. And goodness knows,
there are a lot of tech topics that I cover
in the show where I can't find a reliable source
to give me a definitive date for certain things. The

(10:45):
further back you go, the harder it is to do
that unless someone has specifically filed a document. But in
this case, we do have that. We have an official
filed document. In fact, Chad GBT said that the patent
could be one of the sources that pulled from it.
Obviously wasn't because if they had pulled from that source,

(11:06):
it would have said nineteen fifty two, not nineteen fifty one.
This was a claim that was easily verifiable. The AI
did not do that, But no big deal. Human writers
make these kind of mistakes all the time, right, like
they just don't take the extra effort. Heck, there are
episodes of tech stuff that have gone out there where
I have made similar errors because I did not take

(11:27):
the extra step to actually verify something that's on me.
So I don't want to hold AI to too high
a standard, but it's already proven to not be one
hundred percent reliable, which is an issue if it's being
presented as being one hundred percent reliable. Right not to
say that AI has to be better than humans, but

(11:47):
at least has to be as good as it's claim
to be. Well, what happened next is I would argue
a very big deal and a real issue, more so
than the fudging a year. That is because the AI
creates an expert and it wouldn't be the last time
the spoiler alert, and I'll say it again when we

(12:07):
get to each one, but all of the guest experts
appear to me at least to be invented. I do
not believe doctor Emily Roberts exists. Now. There are plenty
of people named Emily Roberts. There's even one who's named
Emily Roberts who's a child Passenger Safety technician or CPST,

(12:28):
and she creates content for TikTok and YouTube shorts about
the proper way to ensure children are protected while riding
in vehicles. So she's at least in the ballpark for
an expert who's talking about automotive safety features. But there is,
as far as I can determine, no automotive safety historian

(12:48):
named doctor Emily Roberts, at least not anyone I could find. However,
that's probably for the best, right because unless chat GPT
was pulling directly from an actual interview with a real
doctor Emily Roberts, the only conclusion we can draw is
that the AIS fabricating a quote. And I would hate

(13:10):
to see an AI generated script that purported to feature me.
And the words that are being spouted off are not
my own. They're AI generated standing in for me, that
is really troubling. So I do wonder if chat GPT
came up with the name Emily Roberts all on its
own and that's just a coincidence, or if within the

(13:32):
vast database of information it draws from, in the large
language model that feeds into chat GPT, there happens to
be a record of Anne Emily Roberts, who does work
related to automotive safety. Like maybe that was what statistically
guided chat GPT to choose that name. I don't know,
So it's either a coincidence or it's a stone's throwaway

(13:54):
from impersonation, at least in my opinion. Okay, we're just
scratching surface here. We've got a lot more to talk
about as far as this AI generated episode goes. But
before we get to that, let's take a quick break
to thank our sponsors. We're back with our AI generated

(14:22):
technology podcast episode and the issues that I found when
I did a deeper dive. So, the fictional doctor Emily
Roberts explains that early prototypes of airbags used compressed air.
This is true. It is correct at least Hettrick's patent
refers to quote an air accumulator or reservoir end quote

(14:44):
that would hold air under pressure, and then there would
be a valve that would connect this reservoir to deflated
air bags housed somewhere in the vehicle, such as in
the steering column and the dashboard. In Hettrick's design, there
is inside the scent a weight attached to a spring,
and this controls the valve. So normally the valve's position

(15:06):
remains firmly closed because the spring, when it's in its
normal uncompressed state, holds the weight in place. And on
the other end of the weight is the connection to
this valve, which is kind of like a plug. Imagine
a plug in a bathtub, right, and it's connected to
this thing. When a vehicle were to stop very suddenly,

(15:27):
such as in the case as a head on collision,
the momentum the inertia really would mean that the weight
would continue to move forward even as the vehicle itself stopped.
Just as people in the car would continue to move forward, right,
the weight would move forward. This would compress the spring
that otherwise holds the weight in place, and thus the

(15:47):
weight would pull the plug forward this valve, and it
would open the valve between the reservoir of air and
the deflated air bag. So the idea is the compressed
air would then rush up of that's reservoir and into
the area of low pressure, which is the deflated airbag
in other words, and thus inflate it. So this would
be a purely mechanical system. There was no actual censor

(16:10):
to speak of, at least not an electronic one. Momentum
itself would determine if the weight would open the valve
or if everything would stay in place. Now you might
be thinking to yourself, huh, that sounds like it's, you know,
possible for airbags to deploy if you just stop the
car really fast, right, like not a collision, but you
breaked really hard. And yeah, I thought that too. And

(16:35):
my guess is that is a possibility and of a
definite drawback on this particular design. At any rate, Hettrick's
invention didn't really get any traction. It was, as the
fictitious Emily Roberts points out, not reliable enough, so Hettrick
never made any money off this particular idea. It would
serve as kind of a launching point for other inventors,

(16:57):
but it was not sufficient to actually spur the creation
of airbags. But I do have another note about that
nineteen fifty one discrepancy. If you remember, before the break
I mentioned the AI episode said that Hettrick filed for
the patent in nineteen fifty one. He didn't. He filed
for the patent in nineteen fifty two. But there was

(17:19):
someone else who did file a patent for airbags in
nineteen fifty one, a whole year before Hettrick did it.
On October sixth, nineteen fifty one, Walter Linderer applied for
a patent for his quote device for protecting people in
vehicles from injuries in the event of a collision end quote.
Now he filed for this patent in Germany, not the

(17:41):
United States, so it wasn't the US Patent office, it
was the German one. That might be why some sources
cite Hettrick as being the inventor of airbags. It's that
good old American bias, like it doesn't count unless Americans
did it. But like Hettrick's design, Linderer proposed using compressed
air to inflate an air bag in the event of

(18:01):
an accident. Now Lenderers design allowed for either mechanical or
electrical control of the valve that would either allow air
to rush out of the compression chamber or keep it
stoppered up. Interestingly, in Lenderer's invention, there was the potential
for manual control, so in other words, the driver could
throw a lever that would open the valve and deploy

(18:22):
the airbags. I don't know when this would ever be useful,
because you know, when you know how fast a car
accident happens, and how comparatively slowly we humans react to
this sort of stuff. I don't know that there's any
scenario where someone would have the time to switch a
manual lever to deploy airbags and have it be useful.

(18:45):
If anything, I would think such a lever would actually
make it more dangerous to operate a car, because what
if you threw the switch by accident in the middle
of a trip, and then suddenly airbags deployed. That would
not be good. But Lenderer also allowed for automatic deploy
and his proposal was not this mechanical system that Hettrick
had had thought up. His was an electric contact, actually

(19:10):
a pair of contacts that would be built into and
behind the front bumper of the vehicles, so normally the
two points of contact would not touch. There'd be a
gap between them. But if the car were to collide
with something, then the pressure on the bumper would push
back and cause the contacts to touch one another and

(19:30):
thus close a circuit, which Linderer saw as drawing power
from the car's battery to operate the valve electronically and
to have it open. So it was a rudimentary crash detector,
in other words, and one that wouldn't be prone to
the issue of deploying after just a hard breaking you
would have to have contact with something. While Lenderer would

(19:52):
file for his patent in Germany a year before Hettrick
would do the same in the United States, Hettrick actually
got his patent first. He was awarded his patent about
three months before Lenderer would get his in Germany. So
maybe that's why some people think Hettrick is the inventor
of the airbag, because he got his patent first, even

(20:14):
though Lenderer filed his a year earlier. I think all
of this is interesting. I think that this is stuff
that should be included in any episode about the history
of airbags, but sadly we don't get any of that
in the AI generated episode, and rather we seem to
get a case where the host is conflating two different
inventors and the invention date. Now I'm not done with

(20:36):
our fictional historian yet. Doctor Roberts then goes on to
say that the technology becoming more efficient and reliable and
going to solid state is what led to the industry
in general and consumers accepting it. But that's not really true.
The solid state thing, I don't think really plays into
it that much, and Chad GPT doesn't seem to factor

(20:59):
in to account things like the political landscape and the
automotive industry's flex in US politics, and thus the reason
why airbags took so long to become mandated safety features
within vehicles. So there's another cat we got to talk
about who is left out of the AI version of

(21:21):
the episode, and his name is Alan K. Bred, an
engineer and overachiever if ever there was one. So. Breed
was born in nineteen twenty seven, and by the nineteen
sixties he had done more than most folks doing an
entire lifetime. He rushed through a four year college program
in just two years. He did that by attending different

(21:43):
lectures simultaneously, and he managed that by leaving a tape
recorder in one lecture Hall, while he would actually attend
another lecture. So this guy was driven and ambitious anyway.
By the late nineteen sixties, he was tackling automotive safety
and airbag design. The mechanical and electrical switch proposals by
Hettrick and Lenderer were kind of a non starter. They

(22:06):
weren't reliable, and anyway, compressed air would not inflate an
airbag fast enough to provide a cushion for a passenger
in the event of a crash. It would still be
inflating long after the passenger had already completed their full
range of motion in the wake of a crash. So
Breed innovated both on the crash detection part of the
problem as well as the airbag inflation bet Now this

(22:29):
is actually going to overlap into our second fictional guest,
the supposed automotive safety engineer named James Lee. So James
Lee is a pretty darn common name, and there were
lots of hits when I started searching for this to
see if maybe James Lee was actually a real person
that was being quoted here. One James Lee I found

(22:51):
is an engineering manager at a tire company, So you know,
general automotive industry area. There another served as an in
turn for automation controls over at TESLA, So again someone
who's working at least within the industry, though not in
the same capacity as our supposed guest expert. There's also

(23:12):
an associate professor at a tech university who specializes in
manufacturing and mechanical engineering, but not specifically in safety. So
maybe Chad GBT looked around for a likely name for
a safety expert and found some instances of various James
Lees working in jobs relatively in the same industry and
use that name for this episode. Or maybe again it's

(23:34):
just a coincidence. Either way, I feel confident saying that,
like Emily Roberts, there's no real James Lee who is
being quoted here. Instead, we have an invented expert and
invented quotations. So back to Breed and his improvements in
airbag design in the sixties, So on the crash detection front,

(23:54):
Breed proposed a new kind of electronic sensor that folks
would refer to as a ball in tube device. So,
what the heck is a ball in tube? Well, some
of it is just kind of self explanatory. You got yourself.
A small ball made of something like steel, something that's
you know, affected by magnets, and this ball is nestled
inside a tube and the tube is not affected by magnets.

(24:18):
It's made out of something like plastic. And there is
a permanent magnet that holds this ball in place at
one end of the tube. But with enough force, the
ball can be dislodged. It can break free of this
magnetic hold, and at that point it rolls down the
tube and ultimately it hits an electrical trigger that closes
a circuit, so a contact in other words, or it

(24:39):
pushes two points of contact to touch one another. So
in some ways Breed's design married what Hetrick and Linderer
were both working on independently. Now, this is not a
solid state electronic device. Solid state electronics are based on
semiconductors like microchips essentially, So this is an electro mechanical

(25:00):
switch that detects crashes and it and variations of this
would be used in air bags for decades. So modern
cars do have lots of semiconductor microchips, tons of them,
like more than one hundred in your typical automobile, and
these include everything from the chips that connect to entertainment systems,

(25:21):
to navigation systems, to yes safety systems, including air bags.
But the sensors themselves work on mechanical principles involving inertia. Right.
We're not talking about a solid state chip that detects
a crash, not usually anyway. We're talking about this electro
mechanical switch. There are a couple of different variations of this,

(25:42):
so some of them aren't ball and tube systems. Some
use a cam like a little swinging Think of it
like a lever. It's a swinging lever and it can
freely swing, but in the event of an impact, it'll
swing hard enough to push two contacts together and close
a circuit, thus sending the command deploy airbags. Other sensors
use a coiled spring. So think of like one of

(26:05):
those party favors, like the classic party favor, where you
blow into it and it unwraps this long paper tube
that otherwise is coiled up. Same sort of thing, right,
except at the very end of this coil, like in
the interior part of the coil, is a contact, and
when the coil is uncoiled, that contact can touch a

(26:29):
second contact that completes a circuit. So again, normally it's
coiled up, but with a sudden, sudden stop, like an impact,
the momentum will force this to uncoil and thus the
contacts come into touch with one another and boom, you
get your air bag deployed. So again, interesting stuff. But
our safety engineer James Lee and our historian doctor Emily Roberts,

(26:52):
they don't touch on any of this, right, But I
feel like these things understanding how these sensors work is
it's really cool and interesting, but it's not at all
covered in the episode. Also, solid state electronics and automobiles
would become a thing well before the nineteen eighties, which
is what doctor Emily Roberts said was what led to

(27:13):
the adoption of airbags, right, was that solid state crash
detectors were really one of the driving components. I'm using
a lot of puns here, but the AI one did too,
They just didn't acknowledge it, but one of the driving
components that would lead to the adoption of airbags. Solid
state electronics were already a thing well before the nineteen eighties.

(27:33):
And nineteen seventy two article in Popular Electronics by John D.
Drummond details a list of features in upcoming vehicles that
would rely upon solid state electronics and microchips. This was
kind of like the article talking about the automobiles of
the future. Drummond also glumbly pointed out in that article
that vehicles that were most likely to feature this kind

(27:53):
of cutting edge technology would actually come from places like
Japan and Germany rather than the United States, because the
American automotive industry did not have strong collaborative relationships with
electronics companies and semiconductor companies the same way that foreign
car companies did. Like he said, these car companies had
reached out and proactively created these relationships with other companies

(28:16):
that together could collaborate on these kinds of systems. But
in America that wasn't happening. That Detroit was very insular
and they weren't eager to reach out to other companies
and collaborate on this kind of stuff. At any rate,
Solid state components and vehicles were in place well before
we would get up to the nineteen eighties. Even in
safety features now, the AI, James Lee does reveal that

(28:38):
compressed air wasn't the way to go, and Breed agreed
with that back in the nineteen sixties. Lee cites a
chemical compound called sodium azide, and sure enough, that's a
compound found in a lot of airbags, particularly older airbags. However,
the industry has moved away from sodium azide in large
part due to the fact that the compound is both

(28:59):
highly react and extremely toxic. We'll talk more about how
airbags actually work in just a moment, but first let's
take another quick break. Okay, before the break, I was

(29:19):
alluding to the fact that airbags don't rely on compressed air.
That compressed air just isn't a fast enough agent to
inflate an airbag in the event of a crash. So
let's talk about what happens with the chemicals that are
in airbags, such as sodium azide. Well, sodium azide isn't

(29:41):
the only one at the party in these older airbags.
There are other chemicals that are in there too. Some
are meant to accelerate a chemical reaction. So just like
you know, it's like pouring gasoline on a fire, and
it really some of the chemicals are meant to be
the ignition for the chemical reaction, and so literally they're
the most reactive element inside this mix, and they will

(30:07):
then initiate a larger chemical reaction that will be responsible
for inflating the bag. Some of the chemical agents inside
air bags are there to combine with the byproducts of
the chemical reaction that inflates the bag in order to
make those byproducts safer for humans to be around them.
But essentially what happens is you have an electric charge

(30:30):
which ignites the ignition chemicals inside the air bag. They
in turn cause a further reaction, which is essentially an explosion,
and the solid compound rapidly converts to new gaseous components,
inflating the air bag in the process. So with sodium azide,
the explosion creates sodium and nitrogen. Now, sodium is pretty

(30:53):
darn dangerous to humans, is highly reactive stuff, so typically
airbags would also contain something like silica and potassium nitrate.
This would bind with that sodium and thus render it inactive,
so it wouldn't be dangerous to be around. Heaven help
you if that sodium were to come into contact with
water first, that would be disastrous. So this reaction happens

(31:16):
very very quickly. It is explosive, so in less than
thirty milliseconds you'll have an inflated airbag because of this
with the nitrogen gas that's generated filling it. Now, like
I said, a lot of modern airbags actually use other
chemicals to get the same effect as sodium azide. These days,
sodium azide just isn't as common, at least not to

(31:39):
the degree that the AI created James Lee would suggest.
So this is another bone I have to pick with
the AI generated episode. It gives a definitive answer saying
sodium azide is like the most common airbag ingredient today,
and that's not necessarily true. It was true, but it's

(32:01):
not really true now, Like a lot of the industry
has moved away from sodiumazite for various reasons, mostly to
do with safety. So yeah, I take issue with that
particular part of the AI generated episode too. All right,
let's get back to another claim that was made by
doctor Emily Roberts, the automotive industry historian, And she had

(32:21):
said that the industry was quick to adopt airbags once
they became efficient and less bulky, But that's not what
was holding up the automotive industry really. What was holding
them up was that adding in components that don't have
a strong customer demand attached to them and ones that
are not mandated by some sort of government means that

(32:44):
you don't put it into your car because it just
adds to the cost of production and you're not going
to realize a higher profit from using that particular technology.
If no one's asking for it and the government's not
requiring it, it's not going to go in because it
just doesn't make sense from a business perspective. That was
the view of the automotive industry. The automotive industry in

(33:06):
general in the United States long resisted lots of attempts
to impose safety regulations because it would mean things would
get more expensive and it would be hard to pass
that expense on to the customer and plus make even
more profit because customers weren't super excited about safety systems.
That wasn't really what was being marketed to drivers in

(33:28):
the nineteen fifties and nineteen sixties. They wanted cars that
looked fun, fast. They didn't really care if they were safe.
Why incorporate safety devices in your car if you didn't
have to, and if market research didn't show that there
was some sort of demand for it, it's not like
you could use it in order to sell more units,
and if the government didn't give a fig one way

(33:48):
or the other, you might as well just save yourself
the money and leave it out of the design entirely.
That's the real reason airbags would not become a standard
feature for decades, not that folks aren't trying to change that.
In the nineteen sixties, a lawyer who would become a
future politician, Ralph Nader. He wrote a book titled Unsafe
at Any Speed, The designed in Dangers of the American Automobile,

(34:13):
and it really raked the American auto industry over the
coals for ignoring safety concerns and failing to include proper
safety technologies, including basic stuff like seat belts, and generally
the automotive industry contributing to an environment in which thousands
of people were experiencing terrible tragedies, like thousands of people
were dying or getting really badly injured because of traffic

(34:37):
accidents and a lack of safety equipment. To say that
this book was influential is putting it lightly. It was
incredibly popular. It was one of the most popular nonfiction
books on the bestseller list in nineteen sixty six, and
it really pressured the US government to create the Department
of Transportation that year. Later on the US government would

(34:59):
also create the National Highway Traffic Safety Administration in HTSA.
That would be in nineteen seventy. Nator's book singled out
a specific make and model, which was the Chevrolet Corvare,
but his criticisms extended to the automotive industry in general
in America. Now, some states already had laws in place
that required auto manufacturers to include seat belts and vehicles

(35:22):
sold in that state, but it wasn't a US federal
requirement until nineteen sixty eight. A requirement to include airbags
would wait quite a bit longer. It would not officially
become a mandated requirement for passenger vehicles and light trucks
until September one, nineteen ninety eight. That was actually seven

(35:44):
years after the respective law had already been passed. It
was a nineteen ninety one law, but didn't go into
effect until ninety eight. Now, that long gap was partly
there to give manufacturers time to take all the necessary
steps to incorporate airbags into vehicle design. But yeah, the
airbag requirement in the US is actually fairly recent. One

(36:04):
reason it took so long for airbags to become a
required safety feature in the US is that initially they
were being viewed as an alternative to seat belts, not
an augmented complementary safety feature. You weren't necessarily talking about
a car that had both seat belts and air bags,
but rather seat belts or air bags, and auto manufacturers

(36:26):
were able to convince the US government that seat belts
were a much better safety feature than airbags alone, which
you know, not necessarily false, that is true, like it'd
be best to have both of them, But that's not
what the auto manufacturers were focusing on. They were saying, well,
if you're going to require us to have these, at
least make us have the ones that are more effective
and also significantly cheaper, though they didn't say that part

(36:48):
out loud so much so the matter was tabled for
a couple of decades. Airbags just weren't a requirement, but
seat belts were. Now. A few car companies did experiment
with including airbags before it would become required. In the
early nineteen seventies, both GM and Ford would test air
bags in fleets of vehicles, but these were not consumer vehicles.

(37:09):
They weren't passenger vehicles that just any old schmoe could
go buy off a lot. The first passenger vehicle that
was sold to the general public that had air bags
as an option was the nineteen seventy three Oldsmobile Toronado.
GM called it the Air Cushion Restraint System. The Toronado
was a luxury car, so it was not the sort

(37:31):
of thing that, again, an average car buyer would be
in the market for. Like it was way above the
average person's price range for a vehicle, and it would
take quite some time for air bags to become a
more common option in other makes and models. Safety, however,
remained a big concern for Ralph Nader, largely because while
the nineteen sixty eight law required manufacturers to include seat belts,

(37:54):
there was no federal law requiring drivers and passengers to
actually use use them, and apparently like twenty five percent
of folks in cars actually bothered to buckle up and
everyone else didn't. To this day, there is no federal
law in the United States that requires seat belt use. However,
every state in the US, with the exception of New Hampshire,

(38:17):
requires everybody in a vehicle to wear seat belts, and
every state also has laws about seat belts for children.
New Hampshire. If you're an adult, you don't have to
wear a seat belt, but that's that's the only exception.
Penalties for not wearing seat belts vary from state to state.
Some states, the police are allowed to stop someone if

(38:39):
they see that that person's not wearing a seat belt,
but in other states that's not a big enough crime
for a cop to be able to justify pulling someone over.
Now it is a secondary offense, So if the person's
also doing something else, like failing to signal or whatever,
something that is enough for a cop to pull them over,
they can also get cited for not wearing a seat

(39:00):
belt in that particular instance, but they're not allowed to
be pulled over just for not wearing a seat belt
in those particular states. Gradually, more car manufacturers began to
offer vehicles with airbags because it turned out that over time,
the desires of the American buying public were changing and
safety was becoming a bigger concern. So airbags became a

(39:23):
selling point for certain types of cars and the folks
who typically shop for those types of cars. So there
were market forces guiding the automotive industry to start to
embrace the air bag, at least for certain models, but
the industry as a whole continued to resist any efforts
to make air bags a required safety feature. Again, this
was largely because air bags would add to the cost

(39:44):
of manufacturing, and there was a concern that customers wouldn't
be so happy to foot the extra expense, plus you know,
whatever the profit margin the car company had in mind
on top of that. So the airbags issue got battered
around back and forth legally due to various political and
industrial reasons. So in nineteen seventy seven, under the Jimmy
Carter presidency, a rule would have required all car manufacturers

(40:07):
to start including airbags or automatic seat belts starting in
the nineteen eighties. However, in nineteen eighty one, under the
Reagan administration, as part of a widespread attempt to deregulate everything,
this rule was struck down before it could go into effect,
but in nineteen eighty three, the Supreme Court overturned that decision.

(40:29):
In nineteen eighty four, the US passed an amendment to
federal motor safety standards that would require all manufacturers to
include a passive restraint for the driver of any vehicle
that was made after April first, nineteen eighty nine, so
that could either be an airbag or an automatic seat belt,
because again, not everyone would even bother buckling up when

(40:50):
they would get in, so automatic seat belts took that
decision out of their hands. It would automatically end up
buckling someone in. By the way, automatic seat belts were
often cited as being less effective than normal seat belts,
which is a reason why a lot of automotive industry
experts were arguing against it, although another reason is again
automatic seat belt systems tend to be more expensive than

(41:12):
just passive seat belts. The mandatory requirement for airbags really
did not become a thing until a nineteen ninety one
revision to US laws, and again that didn't actually go
into effect until nineteen ninety eight. None of this has
to do with the technology of airbags or their efficiency.
It has everything to do with an industry resisting regulations

(41:34):
in an effort to avoid taking on more costs. So
I feel the AI generated episode does a disservice by
painting an oversimplistic picture of what was going on. But
let's breeze through the rest of this AI generated episode.
So our next guest expert is Lisa Fernandez. She appears
to be another fabricated expert. I guess it's possible that

(41:56):
the former Olympic softball player Lisa Fernandez, who now is
a coach at UCLA, is also really into air bags,
but I couldn't find any information supporting that. So this
presumably fake guest does talk about different kinds of airbags,
and there certainly are all kinds of different airbags, including

(42:17):
things like side impact airbags and curtain air bags. So
this section from a content standpoint is more or less
fine once you get past again an invented expert. Next up,
we have another really important bit, which is safety statistics
and airbags, and this has a supposed expert from the
National Highway Traffic Safety Administration or NHTSA, a statistician named

(42:42):
doctor Michael Harris. Now I found a safety claims adjuster
who goes by Mike Harris. I found a doctor Michael
Harris who is a biological safety scientist and biomedical engineer
at NAMSA. There's an Atlanta Braves center fielder named Michael Harris.
The second, but again we appear to have ours a
fake expert, although maybe Michael Harris the second can talk

(43:04):
to Lisa Fernandez. They're both coming from the world of
sports and apparently getting involved in automotive safety. Again. When
you have a supposed statistician from an official agency providing
actual safety figures, but it turns out that person is invented,
I think that's an enormous ethical problem personally a bulk

(43:25):
at that Anyway, chat GPT's host says that the NHTSA
claims thousands of lives have been saved due to the
introduction of airbags. That checks out. The NHTSA's overview on airbags,
which is available on the web at the NHTSA website,
says quote Frontal airbags have saved more than fifty thousand
lives over a thirty year period end quote. So we're

(43:46):
good there. Next up, this supposed expert cites that airbags
contribute to a reduction in fatalities for frontal collisions for
both drivers and right front passengers. There's a twenty nine
percent reduction in driver fatalities and a thirty two percent
reduction in passenger fatalities. Now I was able to track

(44:08):
these statistics down. They come from a twenty fifteen report,
and it is from the NHTSA. It is titled Lives
Saved by Vehicle Safety Technologies and Associated Federal Motor Vehicle
Safety Standards nineteen sixty to twenty twelve. The information on
driver fatality reductions is on page one twenty five. The
one on right front passengers is on page one twenty seven.

(44:30):
Just in case you want to check up on it
yourselves now, this expert also cites a statistic saying side
air bags with head protection reduced driver fatalities by thirty
seven percent in cars and fifty two percent in SUVs.
This figure seems to come from a totally different document.
This one was written in two thousand and seven by
Anne McCart and Sergei Kirichinko. It's titled Efficacy of side

(44:55):
air Bags and Reducing driver Deaths in driver side car
and Suv collisions. So these authors work was published in
a journal titled Traffic Injury Prevention. This is not an
NHTSA publication, so this particular bit being attributed to an
NHTSA statistician is a bit of an issue. However, Traffic

(45:18):
Injury Prevention is a journal with peer review. I think
they use a single anonymized peer to review papers. I
don't know how rigorous the review processes, but it is
peer reviewed, and I would say that traffic injury Prevention
is probably a reliable source. It's just not the NHTSA,
and I would argue the chat GPT generated interview implies otherwise.

(45:41):
So this information does reflect official studies, though from one
report that's nearly a decade old, and on top of that,
this report actually covers information that spans between nineteen sixty
to twenty twelve, so it's even more dated. And then
the other piece is from a paper that was published
in two thousand and seven. I don't have more recent

(46:02):
reports to pull information from, or at least I didn't
uncover any while I was researching this, But I have
to imagine that the statistics are different today. I'm guessing
they're likely better today than they were when this initial
report was written in the NHTSA. And that's just a
guess on my part. I actually don't have the data
to back it up, but I remain I still have

(46:24):
ethical concerns about the fact that they invented a statistician
representing the NHTSA in this AI generated podcast However, at
least the data that was pulled seems to be more
or less above board, so that's good. Okay, we still
have one more expert to go, and I have a
lot more thoughts to share about this, but we're running along,

(46:46):
so let's take another quick break and we'll be right back. Okay.
So final and yes, fictional guest expert is doctor Sarah Kim.
I looked for a doctor Sarah Kim who is an

(47:08):
automotive safety expert. I could not find one. There are
plenty of Sarah Kim's, obviously, but I could not find
one who would have been an expert that a podcast
host would cite on an episode about airbag safety. But
this expert says that AI and advanced sensors will improve

(47:29):
airbag deployment situations. Right now, you have an airbag micro
controller like a microcomputer essentially that's in charge of this
kind of stuff, and your typical airbag microcomputer system, it's
detecting which seats have occupants, so that way it quote
unquote knows which airbags should be deployed in the event
of a crash. So when it detects a crash, it

(47:51):
sends signals to the appropriate airbags to inflate and then
other ones don't have to. So there are microcomputers that
are part of this. This is those solid state electronics
that doctor Emily Roberts talk to us about at the
beginning of the episode. But these sensors also can gauge
the actual impact. So if it determines that it wasn't

(48:13):
a hard enough stop, like if it wasn't a great
enough impact, then it can prevent the airbags from going off.
So if you have like a little fender bender bump,
then your area airbags aren't going to just spontaneously deploy
in most cases, and that's good because otherwise it can
make things much much worse, like the airbags need to
deploy when the crash is severe, but otherwise probably shouldn't happen.

(48:37):
Doctor Kim makes a pretty safe bet that this is
just going to get better in the future, which I feel.
I feel that that's one of those cop out answers
which I'm guilty of doing myself. It's kind of like
I hate doing it now, but man, I used to
be real bad about this, where I would say, I
guess we'll have to wait and see, or some variation
on that. Occasionally I'll still do it, and I hate

(48:58):
myself every time I do. It's just one of those
trophy kind of things. Now, the fake doctor Kem also
says that airbags will protect people outside the car and
that these are in development. There are cars that are
going to have external air bags. This is kind of
true because as far back as twenty twelve, Volvo was
using an air bag system in which a vehicle detecting

(49:22):
an impact with a pedestrian or a cyclist would deploy
an external airbag. Essentially, it was meant to cush in
a person from hitting the windshield of the Volvo. So
the Volvo's hood or bonnet, if you're across the pond
would lift up close to the windshield and an air
bag would inflate there. And in fact, this feature was

(49:43):
present in the Volvo V forty. This was not just
a prototype, it wasn't a concept. It actually made it
into a production line vehicle. But Volvo did discontinue the
V forty in twenty nineteen, and as far as I
can tell, I haven't seen any other vehicles that actually
have this kind of system in it. But yeah, it
did exist for one line of cars for a while. Now, overall,

(50:07):
I think chat GPT made a pretty mediocre tech podcast episode.
Most of the information wasn't outright wrong or misleading, but
it was incomplete. It didn't tell the whole story, and
in some cases it kind of gave an oversimplification and
misdirect on the story. It also required a lot of

(50:30):
fact checking on my part, which makes my heart go
out to any editors out there whose job is to
read over AI generated articles and make sure that they
are accurate, that they make sense, because as short as
that episode was, it took me hours to go through
all of the different statements and to check them and

(50:50):
make sure that they were accurate or see what was missing.
The creation of fake experts is incredibly concerning to me.
I don't know why why chat gpt made that choice, Like,
why was it necessary to invent experts to whom the
host could talk? Why not just create a pair of

(51:11):
podcast hosts. If you want to have it be a dialogue,
make two fictional hosts who talk to each other and
they just cite information, right, They just say the NHTSA
says that, blah blah blah. Why create people out of
whole cloth to stand in as experts? That is what's
going to get you into trouble. You might remember the

(51:32):
story there was a lawyer who presented arguments that turned
out to be AI generated, and the problem was the
AI created case histories that didn't exist, So the lawyer
was citing precedents that wasn't even real. And when the
judge reviewed it and said, hey, hang on a minute,
none of the cases that you're referencing in this report exist,

(51:56):
the lawyer got into really deep trouble and write so
like they were just relying upon AI being accurate and
not just making stuff up. We see even in this
simple example, that's just not the case. Like, I don't
like the idea of anyone just making up an expert.
I could do that if I wanted to in episodes.

(52:16):
It would make the episodes a lot easier if I
just invented experts who gave me supposedly reliable information. But
it's not ethical. And what's with offuscating the sources of
the information in the first place. Why couldn't it just
tell me where the information originally came from. I probably
should have followed up and asked more specific questions about that,

(52:38):
just to see did it ultimately pull this information from
the primary source. Did it pull the info from the
actual report where the information first appeared, or did it
pull it from some blog post or article that cited
that initial report. My guess is the second, but I
don't know for sure because I didn't follow up, So

(52:59):
I can't reilly fault chat GPT for that yet because
I did not take the steps needed to determine whether
or not it was being lazy quote unquote, but off
these skating the sources just seems weird. Like when I
asked it, what sources did you use? That seems like
a pretty straightforward question that it would be able to
rattle off, like, oh, it was pulled from this and

(53:20):
this and this and this, But that was just not
the case. If you're the kind of person who wants
to be sure that the stuff you're consuming is accurate,
it makes it really much more difficult. If the AI
bot can't give you the sources it's pulling from, it
doesn't make it impossible. Clearly I did it, like I
did all the research. It is a lot more work

(53:42):
on your part, to the point where again, I wonder,
what's the point in having the AI synthesize everything? If
you can't be sure that what you're getting is really
accurate and meaningful. Yeah, I feel like this experience has
reaffirmed a lot of my concerns around generative AI in general.
But that's not to say that it's always going to
be this obtuse. I hope that there will be greater

(54:06):
transparency in the future so that it becomes easier to
make sure that the stuff you're being presented is accurate.
This also will depend upon AI training itself on data
that didn't get generated by other less capable versions of AI.
We talked about that in an earlier episode about how
that kind of process can lead to model collapse, where

(54:29):
a large language model essentially crumbles in on itself because
it's been trained on AI generated material. And the more
that happens, the poorer the quality is of the output. Right,
garbage and garbage out essentially. But yeah, that was an
AI generated episode of a technology podcast and then me

(54:50):
eviscerating it. I do think that generative AI can be
great for thought starters, like I think if I had
used this just in general to make a podcast about
whatever tech topic, and then I went through and actually
researched everything, it could help me structure an episode. It
could give me direction on the things that are really
important that I think is valuable. What I worry though,

(55:13):
is that a lot of people and organizations are skipping that.
They're not really worried about that step. They're just using
AI to generate as much junk as they can in
order to monetize it. To just blast out a massive
amount of content on the internet, post ads against it
and hope that the money rolls in from SEO. That

(55:35):
is dreadful. Like I've never liked content farms ever. I
always hated whenever house Stuff Works would do stuff that
kind of started to tread a little close to content farms.
I'm talking about, like lots of image galleries and slideshows
and puzzles or quizzes that would have lots of page

(55:56):
views added to them. I always felt gross about that.
It wasn't my call, but I really didn't like it.
I much preferred writing high quality articles whenever possible. But yeah,
that's not always the world we live in. And content farms,
while they don't provide much of value in my opinion,

(56:18):
they're getting worse through AI generation. So that's it. That
was this episode being written by AI kinda. I hope
you enjoyed it. I hope you learned something. I hope,
if anything else, this convinces you to not rely too
heavily on generative AI for stuff that you have to create,
using it as again a thought starter or a way

(56:39):
to guide you when you're structuring things. I think that's
perfectly cromulent, as the Simpsons would say, but I don't
think it's a good idea to just lean on that
to do your work for you. It doesn't do a
good enough job. It misses important and interesting things that
I think are absolutely crucial to having a deeper understanding

(57:00):
of the subject matter. And it might tell you some
wrong stuff, like when somebody filed a patent, so keep
that in mind. That's it for this episode of tech Stuff.
I hope all of you are well, and I will
talk to you again really soon. Tech Stuff is an

(57:21):
iHeartRadio production. For more podcasts from iHeartRadio, visit the iHeartRadio app,
Apple Podcasts, or wherever you listen to your favorite shows.