By on October 29, 2019

Apple co-founder Steve Wozniak has officially given up on autonomous vehicles, despite previously being a major proponent of their advancement. “I stepped way back [on] this idea of Level 5. I’ve really given up,” Wozniak at last week’s J.D. Power Auto Revolution conference in Las Vegas. “I don’t even know if that will happen in my lifetime.”

Automotive News reported the quote on Monday, noting that Steve’s tune has changed dramatically from the days where he optimistically saw Apple blazing the trail for advanced driving technologies — something that requires one to venture several years into the past. He’s been harder on the systems more recently, openly expressing his growing doubts since 2017.

“What we’ve done is we’ve misled the public into thinking this car is going to be like a human brain to be able to really figure out new things and say, ‘Here’s something I hadn’t seen before, but I know what’s going on here, and here’s how I should handle it,'” Wozniak explained. “A human can do that.”

While brutal, that’s a rather fair assessment of the industry up until recently. Automakers have spent years aggressively promoting autonomous driving technologies, with several promising Level 5 would be available before 2021. Granted, some of the new driver assistance features are rather impressive. But they’re frequently undone by a little roadway grime making its way onto the sensors, some inclement weather, or a little roadwork.

Meanwhile, companies on the cutting edge of self-driving claim they need more testing in various climates before vehicles can be rolled out for commercial use. Personal AVs are even further out, with many worried they’ll never come.

Waymo CEO John Krafcik is probably heading the company that’s the closest to having a reliable self-driving vehicle right now. But even he has reservations. In 2018, Krafcik said he believed vehicular autonomy will always have “some constraints.”

We’ve seen that happening already. The most successful AVs have been those operating on closed loops or off road (airport shuttles, campus busses, tractors, etc.). Those are likely only a few years away, with autonomous farming/industrial equipment already in operation. Robotic taxis probably won’t manifest until much later but have a legitimate chance, since cars can be isolated to a specific urban environment that can be habitually re-mapped. But they may still have to take days off, depending on the weather, until the systems are truly bulletproof.

Long-haul trucking also might be swallowed up by autonomy in a few years. However, many believe the vehicles will still need a human driver for insurance purposes and to help the vehicle navigate off the highway. That could help delay the presumed slaughter of millions of trucking jobs over the next decade. But passenger vehicles still a giant question mark and are likely to remain so until someone has cracked the full-time-autonomy nut and creates SAE Level 5.

Wozniak said this wouldn’t be such a big problem if motoring were more predictable. “[It would be easier] if we were to modify roads and have certain sections that are well mapped and kept clean of refuse, and nothing unusual happens and there’s no road work,” he said.

Sounds impossible, but it’s just one man’s opinion. And he dated Kathy Griffin, so it’s not like his judgement is perfect.

 

[Image: Viappy/Shutterstock]

Get the latest TTAC e-Newsletter!

Recommended

57 Comments on “Apple Co-founder Claims Autonomous Cars Aren’t Happening...”


  • avatar
    APaGttH

    We are deep in the trough of disillusionment in the Gartner Hype Cycle and heading to the basement.

    I’ve written before, I’ll write it again, Lyft and Uber are basically doomed as they’ll never be profitable to a level to satisfy Wall Street without autonomous vehicles.

    The winners that will emerge on the other end of the hype cycle is likely Tesla, GM, and Nvidia.

    Tesla and its sales aren’t dependent on self-driving technology, but they’re now four years late to the promised delivery of Level IV-V autonomy and nowhere near close to solving the puzzle.

    GM Supercruise is arguably an even better implementation than Tesla Autopilot (rage away, and blinded driver observation cameras aside). That technology will trickle down outside of Cadillac into other models. Next to Tesla, GM is in second place of the OEMs.

    Nvidia will still be selling a lot of processors in this space for years to come, and the development of faster chips will continue. They have the lead in this space and the resources to continue their investment.

    Waymo is part of Alphabet, so they’ll continue to move forward, but the promised technology is still years, if not decades away.

    The truck driver apocolypse is going to have to wait but make no mistake about it, the robots, long-term, are still coming for our jobs and steering wheels.

  • avatar
    thegamper

    I have two 2019 cars, both with a full suite of advanced safety features. On both cars, they dont work consistently enough or as intended such that I would feel anywhere near comfortable letting the car do any of the actual driving for me.

    I tested out the Honda Sensing automatic cruise control to a stop. Pulling up on stopped vehicles from 60 mph is downright terrifying. Far more abrupt braking than pretty much any human driver would be comfortable with. We are still a long ways. I have about 30 more years of driving, hopefully Level 5 ready by then. Will be nice for elderly to still be independent and not danger to the public from behind the wheel.

    • 0 avatar
      ABC-2000

      thegamper;
      “Pulling up on stopped vehicles from 60 mph is downright terrifying”
      This is so incorrect! I drive a 2018 Accord and use ACC every single day, it always behave much smoother than any human driving, including approaching stopped vehicle.

      • 0 avatar
        thegamper

        Well, I have a 2019 Accord…..and its terrifying. If there are humans that brake like that, I dont want to ride with them.

        • 0 avatar
          GoNavy99

          @ thegamper

          I’m with you on this one. 2018 Odyssey and I’ll often apply the brakes before the system does simply because it brakes wayyyyy too late for comfort..

          • 0 avatar
            APaGttH

            My Buick Lacrosse is the same way. If I have the “following distance” set at the middle setting, and traffic in front of me comes to a dead stop, the Buick comes up on that traffic like it’s going to hit at ramming speed. It then applies the brakes in what seems to be way too late, and the last 15 to 20 MPH comes to a gentle stop. It never vibrates the seat or flashes a collision warning on the HUD, but no way would I intentionally brake that way coming up on traffic at a dead stop.

            It also takes it sweet time restarting when the traffic in front of me starts moving. For regular highway cruising, I love the system, it keeps me sane during the morning commute as traffic speeds up and slows down, and in my back-to-back drives from SEA to SMF under less than ideal conditions, the system delivered.

            But coming up on dead stop traffic? Scary experience. If my “brain” can go, “sea of red brake lights, traffic no longer moving,” and initiate a gentle slow down certainly the system could have been programmed to do the same thing. The laser range finder knows the exact distance and closure speed, I’m guesstimating.

      • 0 avatar
        roverv8i

        There are a few things going on here. One, for those that are scared by it I would say it is mostly because you don’t trust it and are over sensitive to what is happening. If you really want to understand just how this can feel try this experiment if you have access to a vehicle of this type. A friend’s mom was a rural mail career and had a Wrangler with full dual controls. My friend and I would do team driving where one person steers and the other person works the pedals. It would never really concern me if I was doing the pedals but if I was steering I would often freak out that he was not going to brake in time. When you are not in control the ordinary can seem extreme. Surely you have ridden with someone driving and you felt like they were not aware of what was happening around them even though there never was an issue.
        Second, there are two types of drivers out there, the one that drives by the seat of their pants and the other that is data driven. A friend of mine is mostly blind in one eye so lacks depth perception. This means you drive very calculating and in her case many would often get concerned that she was not going to stop because they just see something coming at them fast (basically they would have braked way early) but don’t really observe the details what what is going on around them. I never once was concerned when riding with her.
        Third, if the car was really stopping as fast as you think then you would certainly have people behind you that were not on top of there game coming close to hitting you or at least a lot of nose diving behind you. If other drivers don’t seem put off then I am guessing you are not really stopping as fast as you think, you just don’t trust the car.

    • 0 avatar
      sgeffe

      If I try it, I take my 2019 Accord’s speed down to 25mph or less using the brakes, then resume the ACC. When the car is sitting at a stop and I want to get just a little closer to the car in front, I’m thankful that I’ve got my seat belt on because the car clamps the binders on hard enough that it’ll throw an unbelted person forward! After a few months of ownership, I’ve started to learn how to finesse the throttle to prevent the behavior!

  • avatar
    ToolGuy

    “What we’ve done is we’ve misled the public into thinking this car is going to be like a human brain to be able to really figure out new things and say, ‘Here’s something I hadn’t seen before, but I know what’s going on here, and here’s how I should handle it,’” Wozniak explained. “A human can do that.”

    *Some* humans can do that. Sometimes. With some subjects/tasks. Sometimes we are dead wrong.

  • avatar
    dwford

    We’ve been sold a bill of goods on autonomous for years, and now the OEMs are finally tapering expectations. Unless the government bans human driving and forces the switch, I don’t see human drivers and autonomous cars coexisting on the same roads for many many years.

    • 0 avatar
      dal20402

      If autonomous technology becomes good enough to be a serious improvement on human drivers (which I think it will in time, doldrums of recent years notwithstanding), it will be insurance companies, not the government, that force the switch. If we ever get to a point where robot cars cause, say, less than half the claims caused by human-driven cars for an equivalent number of miles, it will quickly become extremely expensive or maybe impossible to insure a human-driven car.

      • 0 avatar
        statikboy

        I think you are right about this, but if AV’s are mandated then the cost of insurance should be pushed to the driver, as it currently is, which means the technology. The cost should be split by the developer of the tech and the ones (insurance companies) who choose to use it. This ruins the insurance companies’ business model.

  • avatar
    pmirp1

    May be Steve realizes how risky these so called self driving cars are and how easily people can mess with them. It is not too hard to find out the various street signs and details lots of these technologies key on, and then mess with them so that the self driving rules get pretty confused. Like the hackers in China who made it such that the Tesla ran in lane with oncoming traffic. Or how much anger there is among ordinary people in Arizona where these vehicles are being tested. Normal people throw things at them and again easy to mess with them. Still, never ever discount people’s desire to make a buck and wall street greed. That will drive this stuff out before prime time, and only Boeing max jet level disasters will cause a step back.

    Still, Steve is too pessimistic, the demand for these things to serve an aging disabled or poor population that can’t afford average new car prices heading toward 50,000 American dollars will push this stuff out soon enough.

    • 0 avatar
      Robotdawn

      No way. AI is a myth with computers designed on the principles they are now. I bet it’s 100 years before true AI, if it’s even possible. People underestimate the amount of non-math computing the brain does just turning on a blinker (which we know is beyond some brains power).

      Autonomous cars I can see working only in the right climate and with the right controlling limitations like restricted routes/highways and restricted conditions. Like a bus, but more private.

      FYI, trucks (as a former truck driver) would be easier to automate. They already are restricted by law as to where they can or can’t drive. I could see the long haul routes being taken over by machines. I can also see fixed route buses and other mass transit solutions being automated.

      • 0 avatar
        GoNavy99

        I studied CS at a quality public US institution. I’m with you here (“100 years before true AI”). “AI” right now is, simply put, a really fast statistics calculator. I’m confident that isn’t how the brain works. I’m in no way afraid that we’ll see any sort of truly disruptive AI in the near future.

      • 0 avatar
        TimK

        ^^^^
        In the 1990s Roger Penrose offered a mathematical argument that showed machines are incapable of mimicking the human brain. Of course he was roundly criticized and savaged by the AI crowd, but here we are 25 years later and all they have are some half-assed systems no one in their right mind (ha ha) trusts. Wozniak is echoing what Penrose said and he has a bit more cred among the geek crowd.

        Penrose’s argument against strong AI is based in part on Gödel’s incompleteness theorems. Digital computers will never be equivalent to the human mind because they cannot process (as Edgar Alan Poe put it) “the thought of a thought”. IOW computers cannot operate abstractly outside the limits of their programming. Whenever self-modifying code is tried, the results are always hilarious (and/or terrifying).

  • avatar

    I think most rational people believe autonomous vehicles were nothing more than hype. Wall Street in a zeal to find the next big thing jumped on the autonomous bandwagon. Like the dot com hype nearly 20 years ago this trend will fade into obscurity. In the end, rationality always prevails.
    The same thing will happen with electric vehicles. I just don’t see EVs getting more than 5% of the market. I have to laugh at Mary Barra’s ridiculous claim that GM will go form selling 16,000 bolts a year to selling 1 million EV’s by 2025. Current statistical analyst just does not back up EV proponents claims that there is a purely EV future in the cards.

    • 0 avatar
      Robotdawn

      To be fair to Mary Barra, she is only saying that to appease those same Wall Street nimrods who think autonomous cars and electric vehicles are the hot commodity. She probably rolls her eyes internally every time she has to say something like that.

      That being said, I would buy a Bolt if it were the same price as a Cruze. But alas.

      • 0 avatar
        dal20402

        Cut battery prices in half from what they are today, and the Bolt would achieve a lower TCO than the Cruze over any reasonable period of time (although still a higher initial purchase price). That’s coming reasonably soon.

        • 0 avatar
          Robotdawn

          I agree in principle, but they’ve been saying “reasonably soon” for quite some time. About 10 years in fact. Is “reasonably soon” 3 years? 10? 20?

          • 0 avatar
            dal20402

            If I had to bet, I’d bet 6-8 years. That’s about the point where a lot of people are forecasting EVs will become the majority of vehicles sold in first Chenese and then European markets.

      • 0 avatar
        highdesertcat

        ” I would buy a Bolt if it were the same price as a Cruze.”

        Did I miss something here? The Cruze retails for ~$17K-~$21K. The Bolt is north of $35K before the tax subsidy.

        • 0 avatar
          SPPPP

          @highdesertcat – Correct, hence the lagging market share.

          @NeilM – Here is an interesting piece on what might have sparked the use of Nimrod as a pejorative (please ignore the saucy title):
          https://dotandline.net/when-bugs-bunny-beat-the-bible/

        • 0 avatar
          dal20402

          You can get a Bolt for $30k or so out the door today, but the difference is still big.

      • 0 avatar
        NeilM

        When did the term “nimrod” become a pejorative term in American English? In the Bible, Nimrod was the great-grandson of Noah, a king and a “mighty hunter.”

        Obligatory on-topic content: Because one day I’ll be no longer able to drive, I’d like to think that Woz is wrong so that my car will drive me where I want to go. But I think he’s probably right, at least in terms of a fully autonomous car that’d go anywhere.

        • 0 avatar
          Thomas Kreutzer

          You can thank Bugs Bunny for the modern usage of Nimrod.

          In the 1940 cartoon “the Wild Hare” he called Elmer Fudd a “poor little Nimrod” as a way of insulting his hunting abilities. But many people didn’t understand the reference and, instead, mistook it as an insult or the equivalent of “idiot.”

    • 0 avatar
      White Shadow

      Electric vehicles are here to stay and the numbers will only increase over time. Anyone who doesn’t realize that surely has his head in the sand.

  • avatar
    SCE to AUX

    Woz is right.

    But the real limiting factor is legal liability, not technology.

    • 0 avatar
      285exp

      Exactly, even if it were shown that overall traffic deaths would be reduced 50%, the lawsuits resulting from the technology failing to prevent crashes, or even causing them, would bankrupt the car makers or their insurers. The families of the people killed by the technology won’t care about all the lives that it saved, just the ones it killed.

  • avatar
    JMII

    The restricted roadways (like GM’s SuperCruise) or restricted lanes (like HOV lanes today) along with between car communication (mesh networking) *could* make this work. But the idea of a truly self driving car just figuring things out is maybe 100 years away.

    Cameras and computers are getting smaller and smarter, but what happens with a little bit of rain or snow? What about construction? I’ve been on roads where they was nothing to indicate where the “lane” was, just random cones. What about when a light is out and some cops are just directing traffic. Or the other day when someone waved me passed them because they trying to get into another lane. Stuff like this is impossible for AI to figure out. Humans can communicate with just a glance, try teaching a computer that.

    • 0 avatar
      APaGttH

      You really think computers need to be 66X more powerful than they are today to achieve self-driving vehicle?

      (serious question)

      • 0 avatar
        JMII

        At a human level yes. The visual processing required seems to beyond our current technology level. We already have self driving cars but calling these “autonomous” is a stretch.

        Per the other article today it seems cars can’t even avoid pedestrians

    • 0 avatar
      SPPPP

      Maybe robot cars will supplant human drivers at some point. But the nationwide rail network is the ultimate restricted roadway, and autonomous technology has not been implemented there yet. If the far easier use case isn’t anywhere near completion, why should we expect the more difficult use case to get solved in the near future?

      • 0 avatar
        TMA1

        Going back to rail and liability, I can’t help but remember we used to have self driving trains on the DC metro system. Then 10 years ago, one train rear ended another, and 9 people were killed. It’s been human operated every since.

  • avatar
    Jeff S

    “If autonomous technology becomes good enough to be a serious improvement on human drivers (which I think it will in time, doldrums of recent years notwithstanding), it will be insurance companies, not the government, that force the switch. If we ever get to a point where robot cars cause, say, less than half the claims caused by human-driven cars for an equivalent number of miles, it will quickly become extremely expensive or maybe impossible to insure a human-driven car.”

    I agree with the above. We will eventually have self-driving cars but not anytime soon.

  • avatar
    Master Baiter

    Took Woz a while to figure out what many of us have known for years.

    Driving takes intelligence. Maybe only 0.1% of the time but that’s enough. And artificial intelligence is really, really hard.

  • avatar
    mcs

    Who needs technology? Science will provide us with tiny little chauffeurs that work cheap:

    https://www.smithsonianmag.com/smart-news/rats-will-drive-tiny-cars-get-froot-loops-180973406/

    • 0 avatar
      Thomas Kreutzer

      Cheap? Have you checked the price of cereal lately? It’s out of control!

      • 0 avatar
        mcs

        @Thomas: I was going to suggest just giving them generic cereal, but the rat might not like it and give you a bad Uber rating.

      • 0 avatar
        Robotdawn

        I buy cereal every week and haven’t really noticed any difference in what I pay for the 3-4 cereals I buy. Of course, I only buy cereal at Walmart.

        I can’t imagine buying brand name cereal elsewhere. The “sales” are usually higher than Walmart prices, and usually are kids cereal most weeks.

  • avatar
    Russycle

    “Robotic taxis probably won’t manifest until much later but have a legitimate chance, since cars can be isolated to a specific urban environment that can be habitually re-mapped. But they may still have to take days off, depending on the weather, ”

    Well, nobody wants to take a taxi when it’s raining, so that should work out fine.

    • 0 avatar
      Tele Vision

      @Russ

      Made me laugh. More pursuant to the point, though, is that the AI can’t just be as good a driver as a human driver is – it has to be better than the best human driver is. We’ve all seen meatsticks make moves on the road and are only alive afterward by dint of the best efforts of those around him or her to avoid an accident. Either all cars are suddenly autonomous ( shudders ) or it’s Meatstick Mayhem.

  • avatar
    ABC-2000

    I recommend watching “Look who’s Driving” on PBS Nova.
    It’s pretty much tells you in great detail why is it so complicated and defiantly not ready any time soon.

  • avatar
    smartascii

    Commercial aircraft pretty much fly themselves. This costs millions of extra dollars and relies on redundant systems, along with two pilots per aircraft and external control of routes, altitudes, and speeds to keep it all working. And as Boeing can tell you, one malfunctioning sensor and poor coding can and will still kill everyone on board. These systems don’t have to contend with traffic in any meaningful way and are maintained and tended by a team of very well-trained professionals. The idea that a $30k car with one to three radars and a camera could hope to safely drive a car on current roads with other drivers is utter nonsense. If you don’t believe me, check out what the most advanced systems out there do when a car in an adjacent lane puts on its blinker (Hint: Nothing, because they can’t reliably identify a turn signal, and they don’t know what it means).

    • 0 avatar

      And the most amazing part is that airplanes live in an easier world for this than cars. No airplanes cutting you off, all signals are remote and received by radio, and airspace is well segmented.

      • 0 avatar
        mcs

        “No airplanes cutting you off”:

        That’s actually happened to me a few times. Also, before tackling automotive collision avoidance, I co-developed aviation collision avoidance systems including one that covers ground/airport situations. There is less to avoid though and even drones are getting ADS-B.

        ” Nothing, because they can’t reliably identify a turn signal, and they don’t know what it means”

        At least one system under development that I’m familiar with can not only pick up the turn signal but actually look for other cues that a lane change is going to happen since you can’t depend on a turn signal. At some point, we’ll have systems better than humans, but it’s going to take a new generation of AI that’s still in its infancy and years of research. There are also sensor advances that will get us there. Like one of my favorites which involves analyzing shadows and using reflections to detect potential hazards. Yeah, the current stuff is crap, but the real thing is on the way.

    • 0 avatar
      MBella

      Also, much like assist systems, aircraft automation is constantly being corrected by the crew

    • 0 avatar
      Super555

      Well we should totally take the drivers out of 80,000 lb semis and trust these systems to work in them. Seems safe!

  • avatar
    28-Cars-Later

    Thank you Woz, I couldn’t agree more.

  • avatar
    MBella

    My latest company car has the full suite of assist systems. The active lane keep assist has yanked the wheel to left on me three times last week. If I wasn’t paying attention, it would have plowed the car into the concrete barriers. These were completely sunny and dry days. This technology just isn’t ready, and Wish is probably right, it won’t for a very long time. The cameras are still hitting the breaks for bumps and plastic bags.

  • avatar
    roverv8i

    I think Steve is mostly on the right track with his thinking. If we were willing to invest in new infrastructure then we would not need level 5. Instead more of virtual track that the car can follow and know where every other vehicle around them is. You might even be able to still allow lesser cars as long as they broadcast certain data and the driver has been licensed for this specifically. Instead we are expecting a vehicle to navigate the current infrastructure coexisting with humans. My thought on this is that if we achieve such a thing then we are very close to being able to have other types of robots and even rudimentary androids. I think people are stuck on the idea of self driving cars and don’t really consider that this is just a special purpose highly advance robot. If we have level 5 cars then I expect we should be close to having a Rosie (the Jettsons robot) too

  • avatar
    Polka King

    Hey Apple. You want to do the world a favor? Do something about traffic light systems. If they can put a man on the moon surely they can make it so there are lines of cars waiting for nothing much less frequently.

    Of course this would involve cameras and scary software and monitoring people, which would be right up Google’s alley.

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • jack4x: I should have specified street legal. But yes, that’s the idea.
  • JMII: His stuff seems mostly like click bait. Thus I assume confusion is part of the strategy.
  • EGSE: Christ on a bike Tim….when the he11 is the “Jetpack has locked your site” bull sh!t going to...
  • SCE to AUX: I was tempted by the Alfa Romeo Stelvio until I drove one.
  • jack4x: As was discussed ad nauseum in yesterday’s Ford decontenting article, most of the features that...

New Car Research

Get a Free Dealer Quote

Staff

  • Contributors

  • Timothy Cain, Canada
  • Matthew Guy, Canada
  • Ronnie Schreiber, United States
  • Bozi Tatarevic, United States
  • Chris Tonn, United States
  • Corey Lewis, United States
  • Mark Baruth, United States