My Top 10 Predictions Are On-Target: Status of AI Self-Driving Cars

Dr. Lance B. Eliot, AI Insider

Image for post
Image for post

I made some predictions about AI self-driving cars for this year of 2019, doing so at the end of 2018. So far, I’m right on-target. Here are those predictions.

— — — — — — —

Achieving true AI self-driving cars will be a crucial evidentiary application of AI for two major reasons.

First, if AI can be developed and deployed for a real-world complex human-based task such as driving a car, it would mean that the field of AI has made tremendous strides and likely a large number of other kinds of hard problems can also be handled via AI. Thus, the driving of a car is merely an exemplar that has forced AI to move forward sufficiently to take on an extremely challenging task that heretofore only cognitively aware humans could do.

Second, the aspect that we would no longer presumably need humans to drive a car and that the driving of a car could entirely be done by an AI system means that the nature of driving and the mobility by society will be radically altered.

The rub is whether we’ll get there without a great deal of torment and hand-wringing, and the other key factor involves the timing of when this will all come to fruition.

For those of you that are familiar with my stance on the substance of AI self-driving cars, I am generally upbeat about the potential for AI self-driving cars to reach true autonomous capabilities, but I am also one of the first to detect and persistently point out the hyperbole that often seems to catch-hold about this emerging innovation.

Here’s then my carefully assembled Top 10 list of predictions about AI self-driving cars for 2019.

Predictions for 2019

Trumpets blare. The secret scroll is unraveled, and Latin words are spoken. It is time for the 2019 prophecies to be cast upon the world.

Well, that being said, here’s indeed my list of 2019 predictions, which is numbered merely for ease of reference and not due to any particular prioritization.

My top 10 AI Trends Insider predictions about AI and AI self-driving cars for 2019 are:

  1. AI Deep Learning Gets Deeper, But No Eureka Breakthroughs in 2019
  2. Expanded Limited Tryouts of AI Self-Driving Cars Valiantly Pushes On For 2019
  3. AI Self-Driving Car Incidents in 2019 Will Be Horridly Inevitable
  4. Media Frenzy In 2019 About AI Self-Driving Cars Wildly Vacillates Hot-and-Cold
  5. Adverse Incidents of AI Self-Driving Cars In 2019 Gets Safety Finally To The Forefront
  6. AI Sensory Devices Emerging In 2019 Invoke Three Words: Smaller, Better, Cheaper
  7. Coopetition of AI Firms and Auto Makers Widens and Deepens in 2019
  8. Federal Regulations on AI Self-Driving Cars Go To Backburner Until 2020 Elections
  9. Bullish 2019 Startups in AI Stoke Market And Many Are In Self-Driving Cars Wheelhouse
  10. AI Innovations Charge Ahead In 2019 And The AI Game-Of-Thrones Rebirth Continues

Now here’s the details and handy references for each prediction:

  1. AI Deep Learning Gets Deeper, But No Eureka Breakthroughs in 2019

In 2019, we’ll have more Deep Learning (DL).

Deep Learning is considered a form of Machine Learning (ML) and referred to as “deep” because it typically makes use of Artificial Neural Networks (ANNs) that exceed the shallower sizing of the ANN’s of prior years, doing so by adding many more layers of artificial neurons and adding many more artificial neurons. Furthermore, this is more feasible to do then it once was because of the advent of hardware related exploits to handle this larger size, including specialized processors and also supercomputing. The cost too of that hardware has come down and made it more readily economical to do the DL training and deployment.

I don’t though see any breathtaking Eureka-style breakthroughs in DL and ML during 2019.

There will be some interesting tweaks to DL and ML during 2019 and there are going to be novelty uses. Will there be a whopper of a new approach to DL and ML that arises in 2019? No.

  1. Expanded Limited Tryouts of AI Self-Driving Cars Valiantly Pushes On For 2019

The auto makers and tech firms are going to continue onward with their limited tryouts of AI self-driving cars on our public roadways in 2019.

These limited tryouts are intended to see if AI self-driving cars can handle the real-world driving tasks as encountered in, well, the real-world. There is only so much that can be done on a closed track or in a proving ground. There is only so much that can be done via simulations. It is believed that the only way to really get AI self-driving cars ready for driving on public roadways is to put them onto public roadways.

Of course, this is kind of a Catch-22. Some say that we should “perfect” AI self-driving cars before putting them onto the public roadways, but the counter-argument is that we’ll never reach that point without actually allowing AI self-driving cars onto public roadways. Yet, if we allow “unproven” AI self-driving cars onto our public roadways and the results are untoward, it could cause the babe to get tossed out with the bath water, namely it could cause the public to reject entirely the AI self-driving car efforts.

It is a bit of a conundrum. Allow AI self-driving cars to drive on our public roads? Yes, because it will provide the kind of experiences needed to successfully drive on the public roads. No, because it endangers the public and could not only cause harm, it could curtail the effort by starting an unstoppable public backlash.

For 2019, the interest in AI self-driving cars is keen enough, along with the ongoing glory and prestige that comes with combining AI with cars, and the salivating billions or trillions of dollars that ultimately can be earned via AI self-driving cars, that the auto makers and tech firms will expand their public tryouts.

I’d wager a bet that this will be done in a somewhat more timid manner than had previously been the case, partially due to the Uber incident in Phoenix in 2018 that led to the death of a pedestrian (along with various other self-driving car related incidents that occurred). These incidents are a helpful wake-up call for many in the AI and high-tech field that had the classic Silicon Valley “fail first, fail fast” mindset — which might be suitable in some circumstances, but in the case of AI self-driving cars that can cause life-or-death results, perhaps the “fail” part should be given a bit more attention and weight than normal.

2. AI Self-Driving Car Incidents in 2019 Will Be Horridly Inevitable

I do not want to be the bearer of bad news. Nor do I want anyone in 2019 to get into harm’s way. Unfortunately, inevitably, there are going to be adverse incidents involving AI self-driving cars on our public roadways in 2019. It is going to happen. Brace yourself.

There will be incidents involving Level 2 self-driving cars, for which a human driver is supposed to be present and attentive and ready to undertake the driving task. I’ve frequently written and spoken about the dangers that the co-sharing of the driving task entails. Humans are likely to become complacent, they are likely to become distracted, they are likely to misunderstand their co-sharing role, and so on. There is a litany of ways in which the man-machine aspects of co-sharing a real-time task can readily go awry.

As Level 3 self-driving cars come into the market, I’ve already predicted this will actually further exacerbate the problem, rather than somehow ameliorating it. This is due to the aspect that the better the AI systems seem to get, the more the human driver will tend toward obviating their duty to aid in driving the car. It is a deathly spiral and we’ll have to see how this plays out (perhaps less badly than I think, or so I hope).

Rather than contending with the AI and human driver co-sharing potential debacle, there are some auto makers and tech firms that are trying to jump instead to the Level 4 and Level 5 of self-driving cars. This involves either radically reducing the human driver requirement or eliminating the need for a human driver entirely.

Why not then abandon the Level 2 and Level 3 altogether, you might be wondering. Because the jury is still out as to whether or not we can really get to the vaunted Level 4 and Level 5, and if so, when we will get there. Meanwhile, it is tempting and possible to continue to add AI and AI-like features to cars that makes them Level 2 and Level 3 ready, and the market hunger is presumably there that is waiting expectantly to get those features. The plentiful hunger needs to be fed, commensurate with the kind of money to be made, and perhaps brand loyalty to be had too. This feeding though carries great risks and the cooks might discover they offered up the wrong meal.

3. Media Frenzy In 2019 About AI Self-Driving Cars Wildly Vacillates Hot-and-Cold

The media, all powerful, the makers and breakers of those both big and small.

There are many in the media that have no idea what an AI self-driving car is, nor the levels of AI self-driving cars, and otherwise consider all so-called “self-driving cars” to be the same. They sometimes call them “robot” cars, which I find misleading because it tends to invoke images of a human-like walking and talking robot that is going to be driving our cars.

Anyway, there is an abundance of what I call “fake news” about AI self-driving cars. It can be due to the reporters that do not know what they are reporting on. It can be due to the reporters having the wool pulled over their eyes, for which they don’t realize that it is taking place. It can sometimes be a herd mentality, involving one reporter that states something and other reporters rush to report the same thing, doing so without taking the time and interest to ferret out the merits of the matter.

To-date, I’d say that the mass media has been relatively favorable toward AI self-driving cars. When something untoward happens, there is a momentary questioning about the advent of AI self-driving cars, but it pretty much loses attention as the ravenous media cycle of new news overtakes the story. The rest of the time the coverage is usually the gee-whiz kind of breathless indication of a glorious future as a result of AI self-driving cars.

During 2019, I believe that the media is going to find itself getting hotter and colder about AI self-driving cars.

The intensity of excitement about AI self-driving cars will heighten at times, partially due to the incremental tryouts of AI self-driving cars, providing more widespread interest and media reporting. At the same time, since I am also anticipating that we’ll have more adverse incidents, the media is going to whipsaw over to the opposition of these public roadway tryouts.

It is going to be a love ’em and hate ’em kind of year for AI self-driving cars by the vacillating mass media.

4. Adverse Incidents of AI Self-Driving Cars In 2019 Gets Safety Finally To The Forefront

Safety has not been as much at the forefront of AI self-driving cars as you might assume. Indeed, I would say that safety in AI overall is a topic that has not been given due attention. The rush towards developing and fielding AI system will ultimately be brushed back when the safety elements become more pronounced.

Similar to how the privacy aspects of data collection and use by the major social media companies was an “unknown” that seemed to suddenly burst onto the scene in 2018 (though insiders knew this was a ticking timebomb), so too will the safety aspects of various kinds of AI systems begin to more overtly rear its head (another ticking timebomb).

I was pleased and honored to participate in a crucial and pioneering AI self-driving car safety summit in 2018 that was sponsored by Velodyne, a major LIDAR vendor. It was encouraging that the AI self-driving car stakeholders are coming together to get on top of the safety issue before it is slammed down upon the industry, which mark my words, it will be.

With my other Top 10 predictions of more AI self-driving car incidents taking place in 2019, and with the media attention and whipsawing, you can be assured that 2019 will force de majeur awareness of safety and AI self-driving cars. Indeed, already some of the auto makers and tech firms making AI self-driving cars have been hiring “Chief Safety Officers” to aid in their efforts, which is one of those it’s-about-time and thank goodness for the realization moment in this industry.

5. AI Sensory Devices Emerging In 2019 Invoke Three Words: Smaller, Better, Cheaper

Most people tend to think only about the sensory devices on AI self-driving cars and often tend to neglect the innards that have to do with the actual driving task. The sensors get all the grandeur. This makes sense since we can all readily see and touch a sensory device, whether it is a camera, a LIDAR device (the cones you often see on top of an AI self-driving car), a radar device, an ultrasonic device, and so on.

Those sensory devices steal the show and on the one hand it is justified since without being able to sense the world outside of the AI self-driving car, there is not much else that could take place. The AI system needs to know where other cars are. Where are the pedestrians? Where is the road? These all require the use of sensory devices.

If the sensory devices are not fast enough, or if they are unable to collect data, or if they are flaky and provide excessive noise in their data collection, it severely undermines the efforts of the AI self-driving car.

Fortunately, there is a lot of push going on about improving the sensory devices. In three words, smaller, better, cheaper is the mantra for 2019. Expect to see advances in the sensor devices. Expect to see improvements in the software that makes use of those sensors.

6. Coopetition of AI Firms and Auto Makers Widens and Deepens in 2019

AI firms want into the self-driving car realm. It’s a fascinating use of AI and one that provides the scale factor that gets the adrenalin pumping. There’s money to be made in them thar hills too. Plus, heaven forbid that the auto makers themselves somehow try to usurp the grand AI powers of the high-tech firms and make a go into the AI arena on their own. How dare they!

Auto makers realize that have to be in the self-driving car realm or else lose their shirts. Auto firm stock prices are both emboldened by announcements about AI self-driving cars and at times are pounded down by such announcements. The marketplace and stock market look at this as a horse race and whichever horse seems to be leading at the moment, it helps that horse and pummels the others (unless the others appear to be gaining on that horse).

You’ve also got the ridesharing firms enmeshed into this mix too. They cannot allow themselves to be left on the side of the road. Auto makers might use the advent of AI self-driving cars to become new defacto ridesharing firms, wiping out the existing lot that had a brief moment in the sun. High tech firms might do the same and wipe out the ridesharing firms.

Often eyeing each other warily, the high-tech firms and the auto makers and the ridesharing firms are all on the same merry-go-round right now. Who will last? Who will get thrown off? Is it better to go it alone or try to join forces in some fashion? Top executives are losing sleep over this, I assure. Go ahead, ask them what keeps them up at night and the answer is going to be AI self-driving cars.

I’ve already predicted and indeed it has been shown that there is a lot of coopetition occurring in this industry. Coopetition is a blending of cooperation and competition.

During 2019, we’ll see more of it. Companies that you never thought might partner will do so. Companies that partnered will at times enjoy the leverage, and some will come to regret it. Not all partnerships are meant to last. Time will tell.

7. Federal Regulations on AI Self-Driving Cars Go To Backburner Until 2020 Elections

Regulators are not quite sure what to do about AI self-driving cars. They don’t want to be overly regulatory, which can get the backlash that they are hampering innovation. High-tech is still the high horse and the public seems to relish the grand new inventions that are being churned out. Imagine if you were an elected official that got labeled as slowing down the advent of AI self-driving cars or maybe stopped it altogether.

This is a tough spot to be in as a regulator. Besides the Luddite charges of impinging on modern-day progress, there is also the hoped-for claim that AI self-driving cars will eliminate or at least reduce the number of lives lost due to car incidents. What regulator can withstand the brunt of not willing to save human lives!

On the other hand, if AI self-driving cars while on our public roadways in 2019 cause damage, injury, or death, you can bet that the public is going to go after regulators that are after-the-fact perceived as being overly lenient. The potential torrent of anger will likely get those regulators booted out, even if they had little to do with the matter or otherwise might suggest that they were relying on the advice of experts. There will likely be a price to be paid, no matter what a regulator might offer as a defense.

There have been ongoing efforts in Congress to try and pass new laws related to AI self-driving cars. This effort has gone back-and-forth and at this writing is still in regulatory limbo. Will it pass in 2019? The odds would seem low. The year of 2019 is the lead-in to the 2020 elections. Does the passage of an AI self-driving car bill warrant the regulators attention in 2019?

If they pass a law that helps promote AI self-driving cars, it might be handy and they could try to use this as a supportive point when running in 2020 for the elections. But, if AI self-driving cars have had recurring incidents in 2019, and the media has turned negative about AI self-driving cars, and if the law so passed was a contributor towards allowing such incidents (or, at least didn’t try to curtail it), those regulators will get tarnished coming into 2020.

The organic approach will more likely be that the AI self-driving car law will continue to bump along and be debated. Unless there is a pressing need to get it passed, it will languish until after the 2020 elections.

The pressing need to get it passed would primarily be if there were recurring incidents involving AI self-driving cars, and in that case the regulators could try to rapidly toughen the law and get it on the books. This might then be a handy tool for the 2020 election campaigns.

9. Bullish 2019 Startups in AI Stoke Market And Many Are In Self-Driving Cars Wheelhouse

I’m bullish about 2019 for AI related startups. I work with a number of high-tech startups as a mentor, and participate too with several startup incubators and accelerators, plus I am a serial entrepreneur in my own right (meaning that I’ve launched, run, and sold several high-tech firms). I also serve as a pitch competition judge, mainly in Silicon Valley and Silicon Beach.

There are a bunch of start-ups that are right now in the concepts stage that will emerge during 2019. Venture capital (VC) and Private Equity (PE) are still eager to find and fund those high-tech start-ups that appear to have something that might be able to hit the ball out of the park.

In one of my 2018 articles, I discussed the nature of AI startups in the AI self-driving car niche. This will continue in 2019 as a hot area to have a startup. Thanks goes to those readers that sent me info about their proposed startups, of which some I’ve now gotten underway aiding them with their formulation.

10. AI Innovations Charge Ahead In 2019 And The AI Game-Of-Thrones Rebirth Continues

I was honored to be the keynote speaker at the 2018 AI World conference in December 2018, during which I laid out some of the innovative topics that I’d be covering in 2019 regarding AI self-driving cars. I believe these will be potential breakthroughs and key upcoming trends:

  • Deep Personalization in AI
  • Emergency-Only AI
  • Game Theory and AI
  • Byzantine Generals Problem and AI
  • Brute Force Algorithms and AI
  • Chess Playing and AI
  • Empathetic Computing
  • Perpetual Computing
  • Big Data and AI: The Case of AI Self-Driving Cars
  • Cobots and AI Self-Driving Cars
  • Crumbling Roadway Infrastructure and AI Self-Driving Cars
  • Anomaly Detection Advances for AI
  • Multi-Party Privacy and AI
  • Hyperlanes versus Bullet Trains and AI Self-Driving Cars
  • Sports Cars and AI Self-Driving Cars
  • System Load Balancing in AI
  • Rewilding of AI Self-Driving Cars
  • Bug Bounty in AI
  • Etc.

Those topics and more are on my plate for 2019.

AI will continue as gangbusters during 2019. We’ll see more adoption of AI systems. We’ll see more advances in AI technologies and techniques. There will be a hint of concern about AI biases, and a large dose of AI safety qualms, for which I am hoping that the AI community can step-up to the plate and appropriately tackle.

For free podcast of this story, visit:

The podcasts are also available on Spotify, iTunes, iHeartRadio, etc.

More info about AI self-driving cars, see:

To follow Lance Eliot on Twitter: @LanceEliot

Copyright 2018 Dr. Lance Eliot

Written by

Dr. Lance B. Eliot is a renowned global expert on AI, Stanford Fellow at Stanford University, was a professor at USC, headed an AI Lab, top exec at a major VC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store