Crash test dummy’s, Congressional dummies and ohh look AI

Introduction and Favorite Pie

Anthony: You are listening to There Auto Be A Law, the Center for Auto Safety Podcast with executive director Michael Brooks, chief engineer Fred Perkins, and hosted by me Anthony Cimino. For over 50 years, the Center for Auto Safety has worked to make cars safer.

Hey listeners, it’s Tuesday, December 2nd and my name’s Anthony and I love pumpkin pie. I’m not afraid to admit it. Really? Yeah. I think I ate half a pumpkin pie over the last few days. Oh, so good.

Michael: So good. Yeah I lean towards chocolate pecan pie, but that’s just me.

Anthony: Okay. Fred, do you have a pie of choice?

Fred: I had plenty of choice, so I guess that’s the same as no choice, but yeah, apple and pumpkin and, it’s all good. Okay. All nutritious.

Anthony: All right. But none of that [00:01:00] has anything to do with auto safety. Don’t eat pie drive.

Fred: I was in Detroit and I managed to survive the Detroit streets, so that was good.

All good news. That’s pretty good.

Anthony: Okay. But if you’re gonna drive, put your pie down and then drive. Okay. This is ridiculous. All right.

Winter Coats and Car Seat Safety

Anthony: Let’s get into something important for the holiday winter, for the winter weather we have here. This is an article from the Tennesseean Nashville, linked from MSN. How do winter coats impact your child’s car seat effectiveness?

According from the article, did you know that winter coats can be harmful to young children sitting in a car seat? I did not know that winter coats can pose some safety risks in the car seat. That’s because heavy coats can interfere with the proper harness fit for children. According to Nitsa, NITSA recommends taking off the outer layer to mitigate the risks of loose fitting harness.

Whooo Michael, when when your kid was young winter coat on, you’re like, ah, I gotta take her jacket off so I fit that car seat [00:02:00] properly.

Michael: Yeah, I mean it’s, we didn’t, I don’t remember us having any super puffy jackets, but I remember having this, just when you have your kid in a normal day, right?

It’s not cold and you’ve adjusted your seat settings, your car seat settings to whatever size your kit is, when winter comes, you’ll notice that, you’re having to readjust the seat settings to go around their jackets. I guess that’s the extent to which I noticed it, but I don’t, I, I think most people’s response to that, and probably mine, I don’t even remember was.

To, simply readjust the seats so that they fit over the larger coat. And that’s apparently not what you should be doing. You should be making sure, essentially use your car’s heater and don’t strap your kid into the vehicle with a very heavy coat on. ’cause it can prevent the protective aspect of the safety seat.

Anthony: This just makes me believe that car child safety seats are a lot more complicated than people [00:03:00] think.

Michael: I think safety’s a lot more complicated than people think in a lot of respects. But in this case, I guess coats are made in all sorts of sites. I’ve seen some pretty puffy coats out there.

So I imagine the puffier and the more your kid looks like the Michelin man or the state puff marshmallow guy, the more you want to ensure that you probably wanna remove that jacket before, before you put them in the car seat.

Fred: Yeah. Forgive my cynicism, but this sounds a lot like.

Perhaps letting better be the enemy of good. I think that your kid is probably a lot better off in a car seat, even with perhaps non-optimal strap locations in the puffy coat rather than not using it at all I don’t think that this reflects the reality of having children, to bring the kid up to the car on a cold day.

10 below zero. Okay, now it’s time to take your clothes off so we could put you in the car seat. I don’t think [00:04:00] that’s gonna work, but I think that there should be an addendum to this study that says, okay, here’s the marginal benefit to be gained from this complicated procedure with your kid. But please don’t let that get in the way of putting your child in an approved car seat.

Yeah,

Michael: I think my takeaway from this is make sure that your child is both warm and safe. And that would be by, when your child is strapped into a car seat, having them, wearing lighter weight clothing and keeping them warm using the incar features that are provided for such,

Anthony: I don’t know.

I didn’t have a car seat. I turned out just fine.

Michael: Yeah I think I was standing up in the front seat of a God, a 1975 Ford Thunderbird when I was involved in a rear end collision, I think I was around four years old. It came out just fine, although some people might argue, I banged my head a little.

Fred: [00:05:00] Maybe my my kids were born in the pre car seat days. Number one came home in a 19, I think it was a 72 Ford Pinto. In the her mother’s lap and a bassinet in the front seat. So all of all of these changes are good, but we need to keep in perspective that, to not let better be the enemy of good on this.

Yeah.

Michael: And part of that’s is motivation behind putting this out is that, I think the last year they have statistics for 20 23, 40 3% of children killed in car crashes. Were not wearing proper restraints. Oh, that’s one reason it’s super important to, to make sure you’re dotting all your i’s and crossing your t’s when you’re traveling with a child.

Anthony: But I, I think this is a call to all sorts of inventors out there is come up with an easier way to properly fit children in car seats. Almost like that. Reebok pump on sneakers. Maybe there’s some way to [00:06:00] do that. You put a kit in and you inflate something so it fits perfectly and they’re strapped in nice and co.

Michael: Sounds like you’re pushing ai.

Anthony: I’m not saying AI at all. No, this is a manual air bladder pump thing. What are you talking anyway?

Michael: Manual air bladder. Like the shoes we used to be able to pump Bam. Like the shoes? Yes,

Anthony: exactly. That’s what I’m saying. Yeah. No, I don’t want AI to do it. ’cause the kid will be like, I can’t breathe.

And then you go, not everybody makes it. You’re expensive. No, sorry, that took a wrong turn. Alright let’s move on to dummies and let this dummy tell you about dummies.

Female Crash Test Dummies

Anthony: The transportation department is endorsing a crash test. Dummy that more cons closely resembles women. Hey, didn’t we talk about this issue three years ago?

Two years ago

Michael: we’ve talked about it a number of times. Yeah, I think we talked about it with Beth Blanks and at least once or twice. Yes. Because we were at the time hoping that Nitsa was going to be pushing through a lot more dummies that, represented the [00:07:00] general population in America a little better.

I’m not sure that this recent news does that so much.

Anthony: It’s a small step. Quoting from this article from NPR women are 73% more likely to be injured in a head-on crash, and they’re 17% more likely to be killed in a car crash than men. The standard car crash dummy using the Nitsa five star vehicle testing was developed in 1978 and was modeled after a five foot 9 170 pound man.

The female dummy is smaller and has a rubber jacket to represent breasts. It’s routinely tested in a, the passenger or backseat, but seldom in the driver’s seat, even though the majority of licensed drivers are women. I got a new dummy that, hopefully will get some better safety results. But Michael, you’re skeptical off the bat.

I can see it.

Michael: It’s mainly because this is the 5% female dummy, so it essentially replicates, a very small woman. Tests using this dummy aren’t [00:08:00] representative of 90% plus of the population of women in America. And, this is the dummy that was, I th I think primarily developed for testing, airbag aggressiveness during the, nineties and early two thousands.

And like you say it’s not a dummy that’s been used in the driver’s seat. It’s, primarily been. Relegated to the passenger seat and rear seats in crash testing. And I think it’s great that it is doing, doing something about this, the current administration is trying to take a lot of credit for moving this ball when in fact this has been under development since, the Obama administration, this particular dummy.

So it’s a very long process. And as we’ve discussed in the past, there’s still a lot of gaps in the data that is underlying the injury metrics that are being used to make determinations about whether, when they’re testing these dummies and crashes, whether they’re experiencing, forces that can injure or kill someone.

There, [00:09:00] there’s still a lot to do there. And in fact, this was just, a preliminary approval by the by, by the current DOT. They still have to issue rules and final rules to, update their anthropometric standards for the dummies. And they have to update their internal crash testing to include these dummies and get them into positions.

And even after they’ve done all that, this dummy still only represents a fraction of the females in the US population. Just and maybe even more so than the average male dummy that we see. The Thor 50 is the male dummy and it, is a five foot 979 pound male which you know, helps you out Anthony, but not Fred and I, not so much.

I’m so happy. You think I’m

Anthony: five foot nine?

Michael: So ultimately, this is a, I think a somewhat significant but very small step in overall, crash testing and I, we still, I [00:10:00] think we could still benefit in the near term while dummies are being developed from, simulation and some other ways of getting at how to protect people in all sorts of sizes.

Anthony: So the male dummy is called the 4 5,

Michael: 4 50, I

Anthony: think it’s four 50. That’s an awesome name. What do they call it? It’s

Michael: the numbers. The numbers. Essentially in the, in those dummy names, define what percentile of the population those represent. So the Thor five represents that fifth percentile female which is very small.

The Thor 50 represents the 50th percentile adult male as of whatever date it was, 19 75, 19 80. I’m not sure the exact date when that was the average size of the male in the United States population. We’re getting slowly taller and definitely wider as the years go on in America. And so there’s a question as to just how well.

These dummies represent the actual humans and cars. And our concern is that when you don’t [00:11:00] have adequate representation from, all shapes and sizes, then only a segment of the population represented by those dummies is actually being protected if cars are being designed based on the responses in those dummies.

So there’s, this is an area that has moved, a lot of things over at the DOT when it comes to rules move incredibly slow, I think crash dummies and crash dummy instrumentation and getting the injury metrics and the data all built and is an incredibly slow process that, that already even, you know and it and when you add onto that, the layers of bureaucracy and the politics involved in decisions that are made in the DOT at nhtsa, that process becomes even slower.

And so it’s disappointing and it, to see this happen and to see the fanfare the administration came out with around it, particularly some of the gender related nonsense they were pushing, which I don’t think I’ve ever seen a DOT secretary do. And certainly it has no relevance to the safety aspect of these of this development was [00:12:00] a little disappointing.

And I just didn’t see any acknowledgement of the fact that they have a very long way to go before we, we were actually seeing dummies that represent all of us.

Fred: No, but it is revelatory to see the Trump administration do anything to enhance the safety of any part of the population. So kudos to them for doing that, even if the advancements actually came during the Obama and Biden administrations.

But yeah, again, long process. If they want to bask in the glory of actually helping somebody, good for them, maybe they can do more of that.

Anthony: Before we move on, I wanna ask, so they, for the five star crash test, they’re using this Thor 50. And does that affect how manufacturers design, for example, the layout of seat belts and how they’re testing airbags impact on the body?

Or is that

Michael: Absolutely. Manufacturers design to the test, they designed to the end cap test, they designed to the [00:13:00] IIHS test because they want those five star ratings. And so our concern is almost always and invariably going to be with, deviations from the test model. Heavier, bigger dummies higher speeds than their testing for.

There are a lot of areas where end cap can be improved from a crashworthiness standpoint, not to mention a lot of the real, the crash avoidance technology and some of the other things that, that nitsa needs to be considering now in crash testing. So overall I’d say there’s a, just a, an enormous amount of improvement that needs to take place here.

And there’s a lot of work that needs to go into it and a lot of research and, that research and that work requires a lot of funding for an agency that’s chronically underfunded. So this is, despite this announcement, this remains a very problematic issue and something that, that I think a lot more effort and money needs to be directed towards.

Fred: And if I can [00:14:00] advocate here for a crash test dummy that is elderly and tall really tall, I, I think that’s what they should do next. Very important.

Anthony: I disagree, Fred. I think every person who drives a car in the US needs to adjust their body to be five foot nine and 170 something pounds, which technically, I think makes you overweight.

But hey, look, it’s America.

Fred: Why don’t you just grow, Anthony, why don’t you just shrink old man?

Anthony: I am, I’m doing that. I’m

Michael: sensing a lack of body positivity on this podcast. We

Anthony: need to stop this. Look, I’m gonna get my leg extensions in. It sounds horrible, but this is a good segue to our next story, Republicans.

The Cost of Safety Features in Cars

Anthony: Say cars are too damn expensive and you know why they’re too damn expensive. ’cause of safety features, not because of things like heated car seats, automatic side mirror defroster all ridiculous edutainment systems that causes [00:15:00] your rear view camera to break. No, it’s safety from an article in the Wall Street Journal.

Republicans on the committee plan to contend that the most effective vehicle safety advancements, seat belts and steps to improve vehicle crash worthiness occurred between the 1960s and 1980s. And that benefits have dwindled since the people said safety technology such as sensors can be ineffective and costly to repair.

They’ll argue. Okay, so let’s, so the Republicans saying all the safety happened in the sixties through the 1980s. Can we identify one Republican during the 1960s and 1980s who said, yay airbags, yay seat belt. I’ll wait here for this. And,

Michael: There it’s possible. I’m sure there were some Republicans back in the day where there was a little more bipartisanship and, conversations across aisle.

I’m sure there were some Republicans who were in favor of some of the airbag and seat belt technology back then. And there probably are still some today, although, I don’t know that their voices are being heard [00:16:00] quite as loudly as, Ted Cruz and my double Gaslight nominee, Bernie Moreno, who is a former car dealer, is, interested in continuing to see these safety systems sold as optional.

This is an interesting hearing for a lot of reasons. First of all, if you look at the. Hearing announcement, it doesn’t mention safety tech whatsoever. It, it seems more focused, it brings up the Electric Vehicle Mandate Cafe Standards program. Ted Cruz is the chair of the Senate Commerce Committee.

And those are two issues that he has been very outspoken on. And, he was a big part of pushing kind of the, what they’re calling now a defacto electric vehicle mandate. I guess that’s an acknowledgement that it didn’t actually exist on the federal level which we’ve been saying for quite some time.

But they’ve eliminated the CAFE standards, which was a horrible thing to do in our opinion. And they’ve eliminated a lot of the emissions. Standards that would’ve protected [00:17:00] Americans and the climate. Ultimately. The only thing they mentioned in, in the release is that start stop technology, which is extremely cheap.

And how that’s contributed to soaring vehicle costs. Given the fact that Bernie Moreno is on this committee, and he recently stalked his fellow senators to try to make a fairly weak point about their adoption of safety systems in their vehicles that we’re going to be hearing in this hearing about the cost of safety systems.

And who better to do, who better to bring in to do that than, the CEO or vice presidents of every major American manufacturer in Tesla and not bringing anyone in to challenge whatever they say. So there, there are no minority witnesses in this hearing, so all you’re going to hear from is the auto industry.

And we know that they’re pushing back on automatic emergency braking as we speak. That’s supposed to come into effect in 2027. I think it may have been delayed somewhat [00:18:00] to maybe 2029 at this point. Those costs have never even, the, they haven’t even shown up yet. So what is making the cost of cars go up so much?

I think it has a lot more to do with inflation, maybe tariffs. Certainly tariffs, I would say. Nothing is free. Safety systems aren’t free. They’re going to cost money, but they’re in, the investment in safety systems produces benefits that go way beyond a lot of the junk that you see being put in cars right now.

You can invest in, you can put heated seats in cars, you can put infotainment systems, you can put, all sorts of creature features and driver assist nonsense into vehicles that doesn’t have any positive impact on safety. You’re passing that charge on to consumers who, many of whom don’t even need that junk to get safely from point A to point B and don’t want it in their cars.

Vehicles are, as we’ve covered many times, getting bigger, can produce a bigger [00:19:00] vehicle, requires more materials that’s gonna raise the cost of cars. So there are a lot of factors behind the increased cost of cars. I think the biggest spike by far occurred in the early days throughout the COVID Pandemic when we saw, semiconductor and supply chain shortages that significantly raised the price of cars.

And we really haven’t seen new cars come down in price since then. And cars in the last decade or so have been, outpacing inflation. And in the past that wasn’t the case. Cars remained relatively, tied to inflation. And so what we’re seeing now is people simply can’t afford cars because of all these broader economic problems.

I don’t think it has anything to do with safety, but in, our current political environment, we’re gonna see, people from one party pushing back on things that the people from the other party did during then their recent period in power. And I think that’s exactly what they’re seeing here.

A [00:20:00] combination of, a reaction to, some of the great things that were done in the last administration, like the automatic emergency breaking rule. And now that the Republicans have power, they are looking to push back on some of those safety regulations, which is one of their, favorite whipping boys.

Otherwise, I think, also I, this is probably a cynical point, but I think it’s worth considering, right? Less vehicle crashes mean less sales, rear end collisions are about, a third of all crashes. I think, you’re probably seeing millions of vehicles every year that are totaled, that otherwise wouldn’t be if we all had effective automatic emergency braking, right?

So the safer the cars get, the fewer replacements are needed. That’s an, that’s a threat to industry sales and industry profits. That may be a very cynical way to look at it, but I, I don’t. I think it’s completely off base, this philosophy of, a, a dealer saying, we apologize for your loss, but you’ve saved [00:21:00] $50 on your next new car.

And, I just don’t think that’s a way to, to promote safety. I, and obviously, and I think we’re scapegoating safety. If it’s brought up, the hearing announcement doesn’t say it’s gonna be brought up. They seem to be focused on the emissions, but the media really believes that they’re going to be looking at safety systems like automatic emergency braking.

And I think that’ll probably occur.

Gaslighting and Safety Regulations

Michael: But suffice to say and this was, also my Gaslight of the week, so I’ll lead into that. I, there’s a couple of gaslights here. The first one is just. The gaslighting, the public to believe that its safety systems and regulations that are actually driving up the price of cars when it’s much more complicated and many more factors involved.

Scapegoating safety and pretending that government regulation is the problem is not going to help any of us out in affording new cars. Removing safety from vehicles is not going to help any of us out in the future. And the other thing is, are they gaslighting us by pretending this hearing is about the [00:22:00] climate and EV regulations when they’re actually bringing in all these manufacturers to pile on to safety systems.

We don’t know yet and we’re gonna have to wait until, mid-January to find that out. But we will be watching and I’m sure reporting what happens and I’m sure we’ll get a few more gaslights out of the testimony we hear on that day.

Fred: Michael, I can’t let you win the. Surprise for most cynical person.

So I, there is a real crisis in rural America of hospitals going out of business. They just don’t have enough, sick and injured people to generate the revenue they need to stay in business. And so the government’s having to step in and help support these hospitals. So if Bernie Marino’s revisions go through, it’s gonna significantly increase the revenue for rural hospitals, which will have a, tremendous spinoff benefit for the rest of the population.

So maybe we need new category of cynicism of the week, but, [00:23:00] there, there’s that too.

Anthony: I think you win it, Fred. You win the cynicism of the week. Thank you. Yeah, but, okay, Fred, I’m gonna call on you. You’re gonna help me ’cause you’re the, you’re really good at math, right?

Fred: Sometimes.

Anthony: Okay. So one of the things in this is that I said is them saying basically all these safety features came out decades ago.

Everything’s fine, right? So I looked and looked at some numbers in here, and I said, I just chose the year 1970, compared it to roughly 2024 when we have data that’s accurate. So in 1970, there was roughly 200 million cars on the road. Today there’s roughly 300 million cars on the road. In 1970, there was 51,000 deaths on the highway.

Car crashes. Today it’s roughly 41,000. In 1970, there was 1.1 million billion vehicle miles driven. Today there’s 3.2 billion mil miles driven. So that’s trillion, trillion trillion, [00:24:00] no billion.

Michael: I think we, we drive about 3 trillion miles a year. Oh, you’d check that? Yeah, we’re in the tease.

Absolutely. Okay. Then it was

Anthony: 1.1 trillion to 3.2 trillion. Okay. But that’s not the important part here. In 1970, there was 51,000 deaths. Today there’s 41,000 deaths. What’s the acceptable amount of death that we’re okay with?

Michael: If you look at the numbers too, we went down, I think to as low as 32,000 deaths per year before.

Going back up which is an argument to me that, you know, and I think it’s backed up by that Mitvah is not doing enough on safety, not doing enough to get technology into vehicles. And there are simply too many delays in getting things in into cars. Roof strength stability control systems.

Even antilock brake systems took a while. Airbags took forever to actually mandate and get into all vehicles. And that’s ultimately what we’re facing here with automatic emergency braking and with some of the other [00:25:00] things that we think are coming down the path that could significantly impact the crash.

Rates like driver monitoring systems that prevent distraction, alcohol detection systems and some things, and speed limiting devices that could seriously impact some of the major contributors to crashes. So all of that stuff is sitting there and some of it is. Technology that’s ready to throw into cars right now.

And the automakers are balking and arguing, oh, we can’t add these costs to the cars. Meanwhile, they’re throwing tens of billions of dollars into autonomous testing, which, is probably not going to be something that drives all of us around for at least another 50 years. And maybe not even then.

So it’s, they have all these other things. They’re throwing money at convenience features, driver assistance systems that don’t have any safety impact. I think that there, you could make the argument that, advertising, there’s a lot of crap that automakers throw money at, that the costs are ultimately being passed down to US.

[00:26:00] Safety systems come so far down the list that it is just almost farcical to claim that safety is behind this increase in car prices.

Anthony: Let me try that one more time. Fred, the question was to Fred. Fred. In 1970, there was 51,000 deaths. Roughly. Today there’s roughly 41,000 deaths. How many deaths are acceptable?

Fred: Me? Yeah, probably none. Yeah. There we go. There we go. You went. But in the spirit of even a blind squirrel, occasionally finding a nut, we need to give Marino credit for saying that the major impact of safety enhancements did occur during the sixties and seventies. There’s no doubt that’s true. The car that you drive today it cannot have a steering wheel shaft puncture your chest because that’s no longer allowed.

The car will [00:27:00] no longer crush your head when it rolls over because of the standards for roof strength. The car, has a seatbelt in it, which will keep you from ejecting yourself through the window of the car and through the windshield of the car when you hit something at 30 miles an hour.

So he’s right in that all those have paused a dramatic reduction in the rate of injuries in cars in the sixties and seventies. But, one of the things that we see from Republicans often is that somebody will observe that a law has worked. So they try to repeal it. We see this with pesticides, we see this with pollution and, people saying that we have a lot of eagles now, so let’s go back to using DDT.

We’ve solved that problem, so let’s recreate the problem. Just, Marino is in the wanted tradition of Republicans, of doing stupid things. [00:28:00] Because the laws that have passed alleviated the problem that they were intended to address. So get rid of the laws and, that’s exactly what’s happening here.

The same tradition that, I don’t know Jim Wright, who’s the last guy who was insecticide or who was a pesticide manager who was a speaker of the house. I can’t remember. One of them was, he wanted to remove the restrictions on using DDT because we’ve got all these eagles now infesting the country.

End of rent. Sorry.

Michael: Yeah. And that’s, that’s arguably already happened with the elimination of the cafe standards and some of the emissions controls that were put into place a few years back.

Anthony: Yeah. Ridiculous. Considering that the auto manufacturers have already spent all this money designing cars to meet these regulations.

So all that investment’s up in smoke. Okay. Fred Michael did his gaslight. Are you ready to do your gaslight? Oh, before we do that, I’m Senator Cruz. I know you’re listening. Please invite Michael Brooks to this little [00:29:00] hearing and he’ll talk and it will be adorable. Alright. Onto to you, Mr. He’s

Fred: available.

He’s available. He’s stand and ready.

Anthony: He’s got a tie and everything.

Fred: So my guest like this week, is a little difficult. It is a book of how to make AI useful by Brian Reamer, who is a research scientist at MIT and Magnus Lint, who’s a futurist, whatever a futurist is. So I was reading this and a couple of interesting points lemme get my to my notes here.

How to make AI useful. We’ll have this link to it in the notes for this podcast. If anybody wants to double this up. There is a lot of useful information here. So he talks about, in a section on AI’s flawed vision of things that happen with ai. So I’ll just read the list. Data theft at scale, that’s interesting.

Environmental strain [00:30:00] due to all the processing, unreliable output, digital pollution, software degradation, catastrophic forgetting, and the opaque nature of ai. That’s a bad idea for driving your vehicle, isn’t it? And then accuracy decay. So following that, another chapter in the book, which is on how to implement it, they go on to say, today’s AI powered driving assistance represent an intermediate step.

For human drivers remain in control, but AI handles specific tasks, reducing fatigue, improving safety, and enhancing the quality and quantity of outcomes. No references are provided. There’s no underlying data or discussion. It appears that they too, have drunk the Kool-Aid or maybe they used AI for their research, so I’m not sure about that.

Is it a gaslight if it’s just wrong, or is it gaslight [00:31:00] require the intent to mislead people? I don’t know about that.

Anthony: It depends what you said. They’re from MIT. It sends what MIT stands for in this case. Yeah. Is it the Miami Institute of Tango? If so, I let it

Michael: slide. That’s, Brian Reamer has actually been a, a pretty.

Pretty outspoken on some safety issues and a and Tesla and some other ones in a very positive way. So I think he’s got a pretty good handle on the subject.

Fred: Yeah, that’s why I was surprised by some of this. And anyway I wondered if perhaps they were using AI to help write the book.

So I went to AI and particularly the Microsoft version of it, it was Ms. Copilot, Microsoft Copilot does, and I asked it, does AI make driving safer? So I did this three times. It’s interesting to see the evolution. So the first answer is yes, AI was already making, driving safer. It doesn’t say safer than what, by the way, by reducing human error, detecting hazards faster, and assisting drivers with real time decision [00:32:00] making.

So the second time I asked exactly the same question, it answered. AI is making, driving safer by reducing human error, detecting hazards faster, and assisting drivers with real time decision making. It is interesting that the first time it said detecting hazards faster and now it’s detect still detecting hazards faster.

There’s a couple other differences though. Studies show AI enabled systems can cut risky driving behaviors nearly in half within months.

AI’s Role in Driving Safety

Fred: That doesn’t show up in any of the other queries, so that’s interesting. And the third time it answered, AI is making, driving safer by reducing human error, predicting hazards faster, assisting drivers of real time decision making.

And that’s an interesting change though. They’ve gone from detecting to predicting now. This seems kinda like inside baseball, but I think it’s very important if you apply [00:33:00] this large language model technology to driving a car, right? Because the cars are stubbornly analog devices, they have a lot of energy.

You have to manage that energy. So if you’re updating your predictions of what the car is doing based upon your detections, those are really different things. And in a complex situation that can change instantly.

Challenges of AI in Real-Time Decision Making

Fred: So how do you triage between the driving directions the AI generated three seconds ago with a situation that it’s seeing now where it’s gone from, one solution to another solution in the same way that this queries went from detecting to predicting?

It seems like a subtle shift, but it’s very important when you’re driving a car to say, I want to turn right. Versus I want to turn left. And if the models are inconsistent, if the processing behind the AI [00:34:00] doesn’t give you the same answer with the same information how reliable can it be in driving a car?

Certain decisions like turn right or turn left are very critical because they can’t be easily changed when you have an analog device like a car that weighs several tons traveling very fast in the proximity of passive of pedestrians. So there’s a fundamental problem with ai if in fact this Microsoft AI is representative of the large language models and that the subtleties and the fine details of what it’s predicting and what it’s directing change moment to moment based upon when you query the data based on the same query or even subtly changing the query.

Because if you’re constantly updating the images. The data that you’re putting into the AI engine to determine whether your car should turn right or left you, you’ve got a real problem. The car just can’t respond [00:35:00] to spurious changes that way. And if you can’t rely on the AI to give you the same answer with the same inputs, how in the world can you ever validate its safety profile?

So it seems like a bit of a rant I get, but I guess, hazards are not risks.

Hazards vs. Risks in AI Analysis

Fred: Further point is that hazards are not the same as risks because for risk you have to attach the probability of it happening. All of the safety references based on ISO 2, 6 2, 6 2 often quoted by the proponents of self-driving cars are all cast in terms of risks, not hazards.

But according to the AI analysis that we just talked about, all they’re talking about at the hazard. Again it seems like inside baseball, but it’s very important in the kind of instructions that are gonna be given to the cars control system based upon the inputs and the rapidly changing inputs coming from the AI engine.

So [00:36:00] that’s already enough. That’s a long gaslight. I don’t know if it’s, and you’re not even sure if it’s a gaslight, right? Yeah,

Anthony: that’s the thing. Your confidence level on gaslight is pretty low.

Michael: I see Anthony knocking

Fred: some points off that, but I thought it was worth, I thought it was worth talking about anyway, because how is

Michael: the book though?

Is it a, I thought it looked like an interesting book because it’s something we talk about a lot, where are the areas where AI can be useful? Obviously there’s an enormous amount of hype in the industry. There are, just like with, when we discussed with autonomy, I don’t think we’re created unleashed on the roads across America, but, if you’re operating a mining company or, off road type, work there, there’s some real use cases that make a lot of sense.

Unstable AI Solutions and Their Implications

Fred: It could be, but again, if the solutions that it’s offering are unstable and what I showed here is that it is unstable. Even with a very simple query, you really question the value of the results. Now, if you’ve got a vast bathtub full of data and you’re trying [00:37:00] to sift through that data.

And the data’s consistent and you’ve got human beings supervising the output to make sure that it’s not stupid. I guess it can be very useful to sift through that data, but I don’t think that the research shows that the results are stable, and I don’t think that the research shows that there’s any tremendous net benefit unless you’re just looking for an excuse to fire people based upon their, substandard performance as predicted by ai, that kind of thing.

It’s useful for that because it can give you a rationale that it reinforces your decision to do something. Anyway, I, so my take, excuse me. My takeaway from the book is that, yeah, there’s a lot of areas where you might apply it, but. Since it is unstable and since it’s opaque and you can’t really tell what the hell it’s doing or why it’s making the decisions, oh, you’re really taking a leap of [00:38:00] faith to use this to manage any complex and in particular safety critical processes.

I think that’s reinforces what Phil Kupin Hass been saying too. It’s just it’s not there. It doesn’t have the stability and it doesn’t have the reliability to allow it to manage safety critical processes on its own. And I think that taking any positive results from a different venue and saying that means it’s safe to drive a car, is a really hazardous and unjustifiable migration of the of the thought process.

Michael: Yeah. And my question that I keep running by my head right now is what would you get in a car that has the potential to catastrophically forget how to break? And that just sounds insane. Why would you put yourself at risk of such a thing?

Anthony: Because I got a app man, and I got points for each mile.

I go, I just asked the AI right now and said, what should I do next? And it said, go to auto safety.org and click on donate. That’s right. It predicts you’ll be a [00:39:00] better person if you do this. No catastrophic failure, no opaqueness auto safety.org. That click donate. That is useful.

Fred: AI giving Tuesday, isn’t it?

Isn’t it official that people should empty their wallets today?

Anthony: Yeah, but people won’t hear this till Thursday. Oh, damn

Fred: it.

Anthony: I know, but okay, get in your time machine. Go back in time. ’cause AI says time travel’s real. Okay. On from my gaslight. My gaslight, I haven’t even shared with you guys at all. Ready? My gaslight’s a good one.

My Gaslight is future friend of the show. Jim Farley. That’s right. The CEO of Ford Motor Company whose job is to get you the consumer to buy Ford vehicles. What’s his favorite car? Xmi. That’s right. A Jim Farley drives a Xmi, SU seven and he loves it. He says, we flew one from Shanghai to Chicago and I’ve been driving it for six months now, and I don’t want to give it up.

Jim Farley driving a [00:40:00] Chinese electric vehicle that you as a normal American consumer cannot get at all.

Michael: Yeah. And Ford tough if you listen to the podcast back in about six months ago, you would’ve known that we covered that.

Anthony: Damnit Steve, Hey, how did, how does

Michael: that work? Did he smuggle it in?

There was a band the correlation. You did an tell me. No that’s the question I think I asked when we covered the first time was how did you get the, because you have to apply for an exemption with the federal government to bring a vehicle in that doesn’t meet federal motor vehicle safety standards.

Didn’t cover this. I’m sure Jim and his lawyers know how to do that real well. So that this, I’m assuming that’s what happened.

Fred: Another example of rich people being able to ignore the law.

Anthony: You know

Fred: what this is that never happens

Anthony: Of me messing up Gaslight. But you know what, nobody wins. Okay.

’cause Fred just went on so long, I don’t even remember what Michael’s was. Okay. Yeah. I’m making sugar gaslight, so nobody wins. I’m taking my ball

Michael: going home. I’m pinning a blue ribbon on my chest today.

Anthony: Okay. That’s a little weird. I won. [00:41:00] Okay.

Subscription Fees for Car Features

Anthony: Hey, let’s go into something that’s near and dear to me. The State of New York.

That’s right. There’s a New York Senate, state Senate bill that’s been proposed put forth, which would make it illegal for manufacturers to charge subscription fees, features already installed on vehicles at the time of purchase. If signed into law, a situation like BMW’s infamous heated seat subscription will be subject to a fine of up to $250 for each sale, which I think is missing a couple zeros from the New York State Senator James Kofu, who’s sponsoring the bowl.

If the automaker is building a feature into a car for as long as cars have been invented, you’ve never had to pay a subscription to access those features. When asked if he seized this as a next wave of corporate nickel and diming, turning safety into a luxury kofu didn’t hesitate. It is, there’s no question.

He gets my hero of the week. I’ve just a minute hero of the

Michael: week. Yeah that’s, it’s that’s something I think we totally agree with. It’s an [00:42:00] interesting bill. We haven’t done a gigantic full analysis of it, but I think the article points out that the interesting part are the exceptions to it, so it wouldn’t apply to your navigation system updates.

I know that’s a pain for people who have to pay a few hundred dollars to update their incar navigation every few years. If you’re upset about that, I’d recommend using an online version and not even dealing with your incar navigation infotainment features. I don’t think I would have necessarily a problem with, people having to subscribe to infotainment features while they’re in their car.

If anything, that’s a barrier to them using them while they’re driving. Satellite radio, that’s typically a subscription thing. And vehicle wifi, I could go with that. But then it says telematics services, which I have a problem with that. That could apply to a lot of things that could apply to post crash notification to the EE emergency authorities in your area.

That’s certainly not something we want to be subscribed to. [00:43:00] Telematics services could also be applied to vehicle or vehicle to infrastructure tech, which would, be awful if that was a subscription service since it’s presumably being deployed to protect everyone in connected vehicles.

And, roadside assistance is one thing, but roadside emergency would be a completely different thing if there’s a crash. I think the bill has a few kinks that need to be worked out. That may be, kind overly kind to some of the manufacturers. Another one that’s pretty obvious, there’s software dependent driver assistance, or other driver automation features.

If there is a driver automation feature that is a safety feature, then certainly we wouldn’t want that being subject to a subscription. So I’m not sure about that exemption either. So I, I think it’s a good idea. But I think there needs to be some more guardrails put around the bill to make sure that manufacturers aren’t able to take this language and run and start charging us for potentially safety related [00:44:00] technologies.

Anthony: With that, I think it’s time to jump into the towel of Fred. How are you feeling, Fred? We’re ready. Ready to go? He said we’re ready. How many people are there inside your head? Oh, 13. Wasn’t Sybil’s record 13 or something like that? But you’re gonna continue with the man, why can’t I remember the name of it?

Regulatory Checklist for Autonomous Vehicles

Fred: Yeah, the checklist for regulators.

Anthony: Checklist. Regulator. Checklist. Computer

Fred: driver.

Anthony: Yeah. Yes.

Fred: There we

Anthony: go.

Fred: Again, this is for context. This is a list intended to fill the gap between the expert information that is really required to interpret applications, to run an AI within your jurisdiction, and the heavily weighted information that you’re receiving from the applicant who says, everything is fine.

We have end-to-end ai. So just shut up and give us approval. So the next one on our list, which is on our website as well, so you [00:45:00] can find that if you want, is has the applicant or AV operations included adequate reasonable physical safeguards and warnings to alert and promote self-defense of motorists and other road users outside of the computer driven vehicle?

If the computer driven vehicle exhibits anomalous behavior or if collision is likely impermanent, what’s this all about? If the car is gonna kill you, you have a right to self-defense and you should know that it as a bare minimum that something really hazardous is approaching you. Now in, in industry, typically they put up physical barriers to keep you away from, oh, I don’t know, steam shovels and lifts and things like that, which as you walk along the street and observe a building being built, right?

So if stuff doesn’t drop in your head. The AV industry wants to proceed without having any such safeguards. So I think this is something regulators need to look into, [00:46:00] whether or not the industry is providing adequate safeguards that are associated with the dangerous operation of the avs. Next up

Michael: is, it’s interesting, I wanted to ask about that real quick here.

There ha there’s some sort of presumption going on that a computer driven vehicle that is presumably operating unsafely actually has knowledge that its anomalous behavior is taking place and can warn other drivers, is that overthinking it? But it just seems like it presumes that, that those physical safeguards can be put in place because the vehicle is aware of the, is aware of a risk.

Fred: Think of a, think of the Waymo approaching the airport in Phoenix, right? It’s dropping off a passenger, but in the process of dropping off a passenger, [00:47:00] it’s encountering, I don’t know, a hundred pedestrians on the sidewalk at the area where people are delivering, where the taxi are dropping people off.

So let’s say that it’s coming in, it knows where it is, but there’s a defect in the speed regulation. It’s coming in at 30 miles an hour rather than three miles per hour, which would be a safe speed. The car should be able to sense that. It should be able to know that it’s proceeding hazardously in an area where there are pedestrians.

And if it can’t slow down, then it should start to beep and holler and blink lights and all that sort of stuff so people can get the hell out of the way. So that’s just one example. Is it possible to do that? We know that Tesla switches off the. Autopilot technology before it crashes into things, so that there’s no record that it was on autopilot when it crashed into things. We’ve seen that [00:48:00] in in the analysis of data that came out of court cases. So we know that at least in that one case, and this is probably primitive self-driving technology, that yes, the car is sensing that it’s in trouble and yes, it takes some actions based on that, that, that make sense?

So the next one up, next item up is, has the applicant shown that it will not endanger or injure anyone by, its near term post collision actions? So if there’s an, if there’s a crash, right? CM crews, what happens after the crash? Does it still try to drive away? Does it, have electric short circuits that are gonna endanger the first responders?

Is there, these are things staff, the lead

Anthony: footage and light of regulators?

Fred: Yeah. There you go. Next up, has the applicant proven the safety of occupants and other road users in the event of electrical power loss? We know that this happens in cars, right? We know that there are records of [00:49:00] people trying to rescue passengers of burning AV vehicles that they, that can’t because the electric power within the vehicles has been compromised and it, so people are dying inside of the cars while other people are trying to open the doors.

This is this is in my mind, unacceptable.

Anthony: I gotta jump in there real quick ’cause that makes me think back to this Senate hearing that’s coming up next month. I wonder if they consider manual door releases in these vehicles. Is that an unnecessary safety feature?

Michael: Since it’s not mandated, it’s not even a, it’s not even on their radar, yet.

I I guess if there was a rule which we would support in development to standardize manual releases on doors to making more, intuitive for consumers and anyone getting into the car then that could be on their list. Although, I don’t know if, I don’t know if the argument, you’re not really adding anything to the car [00:50:00] there, the, I don’t know if the cost argument would really fly.

So

Anthony: regulations are bad. Yeah.

Fred: Yeah.

Anthony: That’s, sorry, Fred.

Fred: Yeah. No, but that, in the absence of regulations, it’s up to the regulators at the local or state level to address these questions. They, and it’s important that they at least understand that the questions exist and how. The developer’s gonna respond to it.

The alternative is to just close your eyes and hope for the best. And don’t worry too much about your constituents. I think that’s an unacceptable position. But until the regulators have a tool in hand that can let them know what the issues are they’re really unable to do anything except rely on the experts.

And the experts are all in the pay of the automotive industry. How’s that gonna work out? How’s it been working out? We see it in Texas. We see it in San Francisco. People being dragged under the avs. [00:51:00] We see it everywhere. And I don’t think this is due to the, any kind of malice on the part of the regulators accepting the past general counsel of Waymo.

I think he was one of those guys who regulate, who was in California. But anyway, assuming the goodwill of the regulators. They need to understand, they need to have the tools in hand to allow them to ask the pertinent questions about the developers and what is being introduced onto the roads. That’s what this is all about.

So the next one up is, has the applicant proven the safety of occupants and other road users in the event of loss of external communications or connectivity, particularly with regard to notification and controlled response to emergency operational design domain restrictions? So what is this all about?

All of the AV companies have remote supervisors for the car in case of an emergency [00:52:00] or in case of a, a car stopping in the middle of an intersection to get it to driveway or any kind of stupefy combination of need and data that causes the car to stop running. So they’ve all got remote.

Links that allow people to tie into the car’s communication system and talk with the occupants, control the car, give it new directions. Those all rely on external communication links. Those all rely on cell phones, right? So what happens in the car if those communication links don’t work? Is there anybody here who is always had a hundred percent communication with your cell phone?

Yeah. I don’t see any hands raising up, but yeah, no. So that, that observation of everybody in the world was ever used a cell phone. Is it particularly important when that cell phone communication addresses the safety [00:53:00] of a stubbornly analog, dangerous moving vehicle?

Anthony: All right. Yeah. Can we can I cut you off just for time?

Fred: Sure, yeah. Continues more next week. Yeah.

Anthony: Great. This is really good. Before we jump into recalls I wanna do a teaser of a subject that we’re gonna get to hopefully before the end of the year. Maybe it’s Waymo terms of service. Now, Fred mentioned this to us before the start of an episode once, and I glanced at it, and I’m just gonna highlight one little piece early on.

This is how insane their terms of service are. This is under heading four, sharing Your Ride Subsection a. Guest Compliance. You are responsible for ensuring your guests receive notice of Waymo’s privacy policy and agree to be bound by these terms, including all terms applicable to you. Waymo saying, guess what?

You’re gonna go to law school. You’re gonna, you’re gonna spend three years in law school before you get into our car, and then you’re gonna bore the shit out of your friends and make sure they have a base level understanding of [00:54:00] the law as well. Enjoy your ride. The rest of their terms of service is just as lunatic, insane.

Nonsense is this, and I wanna do a deep dive on all of this in the future episode, but for now it’s recalls.

Recalls and Safety Concerns

Anthony: Alright. First up, Nissan 41,797 vehicles, the 2025 Nissan Sentra. It’s under glazing materials. Visible air bubbles may be present in the laminate layer of the front windshield on certain Nissan Central vehicles.

Ooh this makes it so the vehicle may not comply with F-M-B-S-S 2 0 5 glazing materials and some other things around testing standards. Oh, that’s a bad one. Yeah.

Michael: Yeah, that’s a bad one. It looks like this is just an issue. The bubbles are essentially going to make the light coming in through your window in certain conditions.

Glare or just, essentially remove your ability to see properly [00:55:00] through your windshield, which is obviously a huge problem. And it looks all they’re gonna be remedying this in mid-January or so, and owners will. I think it looks like owners are gonna bring your car in first. They’re gonna inspect it to see if it has the air bubbles, and then if so, they’re gonna replace the windshield.

Anthony: I don’t know, I think Senator Cruz might think that a clear invisible windshield is an unnecessary safety feature that’s causing too much expense. Next up Hyundai one 143,472 vehicles 2024 to 2025 Hyundai Santa Fe, the Santa Fe hybrid. A malfunctioning or an inoperative, goddammit. Rear view camera will reduce rearward visibility when driving in reverse, blah.

That camera’s main wire harness was in, probably assembled by the supplier ’cause it’s Hyundai. So the supplier was probably a 9-year-old kid causing increased tension in wearing the cable shield line. You had no

Fred: that with fairness to Hyundai, there are no, [00:56:00] there’s no evidence that people are younger than 12.

Anthony: Allegedly.

Michael: Yeah. This one, it, this one is, they’re blaming on the supplier. I, Hyundai is pretty close, works pretty closely with its suppliers, so I’m always, I always wonder when manufacturers are blaming their suppliers in the recall report. Just how you know. Yeah.

Accurate at all. That is, it’s hard to know. But it looks like owners are going to be getting a remedy here in mid to late January. And I think they’re just gonna replace your rear view camera with one that has a better situated wiring harness That’s not going to fail.

Anthony: Next up Hyundai, 85,043 vehicles.

The 2020 to 2023 Hyundai Sonata. A fuel tank assembly in the subject vehicles may become pressurized with compressed air due to a worn purge control system check valve causing the tank to deform information of the fuel tank can lead to fuel leakage over time. [00:57:00] Wait, what how did air get in there?

Fred: Air gets in whenever you fill up your tank with gas.

Anthony: I thought there’s a

Fred: check valve

Michael: that’s supposed to prevent some of this from happening. That looks like the it’s a purge control system, which I’m assuming is supposed to vent a tank, but it looks like they’re trying to go for a software remedy here.

And it’s very confusing. There’s a long. Long description of the problem. And I believe the Kia K fives that are in the next recall are also impacted by this. But essentially, I think we’ve discussed it before where we saw some fuel tanks swelling. And this appears to be what’s happening, pressurized air is being forced into the fuel tanks and that’s causing possibility of a leak due to, cracks in the fuel tank.

Plus, you really have to wonder if that’s safe to drive around with an inflated fuel tank. Nothing about this sounds safe. In fact, it’s a [00:58:00] big yikes from us. And I, in, I, I’m still a little skeptical that the software, this is one of those recalls where software is proposed as a solution and I know, I don’t know that it is the solution.

Maybe it’s the check valve that needs to be replaced. It doesn’t appear to be that the tanks would be replaced, but at any rate, owners are going to have a remedy it looks like by mid-January available. So be sure to get this one fixed quickly because it does not look like something that you want to be messing with.

Anthony: Okay. Fred, I have a dumb physics question. Okay. I’ve got a fuel tank and I’ve maximized it. It’s full of fuel and as the fuel burns off gets sucked out, I imagine air fills that gap where fuel used to be, correct? Yep. Okay. And but you said when I’m fuel filling the fuel tank, I imagine that air’s getting pushed out by the fuel.

Fred: Yeah, it is coming back out through the nozzle. That’s why they have, gas recovery technology [00:59:00] built into the the fuel dispenser at the gas pump.

Anthony: Okay. So when, oh, when I’m putting that fuel in there, it’s pushing the air that’s getting pushed out goes back through that nozzle.

Fred: And it’s okay. And it’s also mixing with the gas that’s going in, so the bubbles are getting into the gas.

So there’s a lot of exchange going on. Do I want bubbles in my gas? You have bubbles in your gas when it goes into the tank. Yeah. And your tank is built with an overflow mechanism so that the there’s room for the fuel to expand without causing any hazard once you filled up the tank. So if you fill it all the way to the top so it’s dribbling out you’ve done a bad thing because you’ve filled up the space that is allowed for expansion of the fuel.

Another thing that happens, of course if the fuel is cold, it’s in the ground and you fill up your gas tank and then you warm it up as you drive, which is inevitable, it’s going to [01:00:00] expand, right? So that’s another issue. So there’s a lot going on there. And as you’ve observed, as you use it up, you’ve got to replenish the air that’s in the fuel tank, otherwise it’s going to collapse.

Anthony: So when I go to New Jersey and they fill the tank for me, nine times outta 10, the guy, the fuel filter will automatically shut off when it’s full, but these guys keep going for a little bit more. I should get out of the car and be like, Hey, don’t you do that?

Fred: Yes, you can do that, but it’s New Jersey, so

Anthony: I don’t wanna do that,

Okay. Yeah, good point. Yeah, they should just weird.

Fred: They should just shut it off and give you the bill when it automatically shuts itself off.

Michael: Okay. Yeah. And this recall is a little weird, interesting because this was a, Hyundai apparently did a, an emissions related service campaign that put in software because it was discovering the fuel links and it was trying to address them.

And from an [01:01:00] emissions standpoint rather than a safety standpoint, they put in software to help prevent wear on the the check valve. And it looks like, even with that software put in, they started seeing some significant safety concerns. And so we’re forced to take an actual safety recall to address it.

And it looks as part of this new safety recall, they will be replacing that check valve as well, not just a software recall.

Anthony: Last recall completing our Korean trifecta is KIA 250. And that’s the same thousand 547 vehicles. The 2021. 2024 Kia K five. Heard just exactly what we just said here.

The valve check valve, assembly line, same thing. Yeah.

Michael: The Kia report doesn’t dive into it nearly as deeply as the Hyundai one, they’re still trying to pretend that they’re separate companies.

Anthony: Alright. And

Fred: yeah, the only difference between a Hyundai and a Kia is the label that they stick on the outside of the car.

Same [01:02:00] factory, same technology. It’s they’re identical. Identical twins, maybe Irish twins, but still, nearly identical.

Anthony: Perfect. I don’t think they’re Irish.

Fred: Could be. I don’t know. I you’re, are you an expert on ethnology?

Anthony: Hey, that’s a, my different podcast. Hi, Jen. I can’t think of a clever name for it right now.

Conclusion and Final Thoughts

Anthony: Thank goodness that’s the end of our show. Thanks for listening to everybody, and thanks more importantly for donating. And if you’re a fan of F1. Oh my God, what is gonna happen this weekend? It’s so crazy. Nobody cares. I do. And friend of the show, Jonathan Gman. Yeah,

Fred: thank you. Bye bye. Thanks. Bye bye.

Bye. For more information, visit www.auto safety.org.