The problem with self-driving cars

We’re excited to bring Transform 2022 back to life on July 19th and virtually July 20-28. Join AI and data leaders for sensible conversations and exciting networking opportunities. Register today!

Alias: Why this burning pit of money has failed to bring meaningful results for decades.

The future is here, and it doesn’t look like much. As we approach Alexnett’s 10-year anniversary, we need to critically examine the successes and failures of machine learning.

We are watching from the high plateau.

We’ve achieved things in computer vision, natural language processing and speech recognition that were unimaginable a few years ago. By all accounts, the accuracy of our AI systems has exceeded the wildest fantasies of the past.

And yet, that’s not enough.

We were wrong about the future. Every prediction about self-driving cars is wrong. We are not living in the future of autonomous cyborgs, and something else has come to the fore.

Growth on automation.

Humans crave control. It is one of our deepest, most instinctive desires. There is no world where we leave it. One of the biggest misconceptions of the AI ​​community today is that people become more comfortable with automation over time. As the reliability of automated solutions proves, the microwave background comfort of society is constantly increasing.

This is wrong.

The history of technology is not the history of automation. It is a history of control and abstraction. We are tool-builders, so obsessed with experiences beyond our control that over thousands of years we have developed entire cultures and myths around the movement of heaven. So it is with all technology.

And so it is with AI.

From the early days, the problem with self-driving cars was obvious: there was no control. When we look at the successful implementation of self-driving cars – many years old now – we look for lane assistance and parallel parking. We look at situations and use cases where the control panel between human and machine is clear. In all other situations, where the goal is to pursue the mythical level 5 autonomy, the self-driving car has failed miserably.

Technology is not a hindrance.

In 1925 we had a radio-controlled car navigating without a driver behind the wheel in a busy traffic jam on the streets of New York City. At the 1939 World’s Fair, Norman Gades’s Futurama demonstration outlined a rational smart highway system that would effectively use magnetic spikes – such as electromagnetic fiducials – embedded in the road to guide cars. He Predicted that by 1960 autonomous cars would be the dominant form of transportation.

Of course, that was also wrong.

Although not about technology. No, “smart highways” have been hugely successful and straightforward where they have been implemented. Even without additional infrastructure, today we have self-driving cars that are capable of driving as safely as humans. However, with more than $ 80 billion flowing into the region from 2014 to 2017, we do not have self-driving cars. For reference, the $ 108 billion committed by the US federal government to public transit over a 5-year period was by far the largest investment in public transport by the country.

Of course, the difference is that I really can Ride Train.

The problem, basically, is that no one has bothered to think about the new control panel that we’re trying to enable. The question was never about automatic driving. It is a myopic, closed-minded thinking. The question is how to transform the transport experience.

Car suck.

They are big, loud, smelly and basically the most inefficient form of transport that anyone can imagine. They are the most expensive thing a person owns after their home, but they are not Create Value is not someone’s property Wants For ownership, it is a property that people have Have For ownership. It is a regressive tax that destroys the planet and subsidizes the highways that damage our cities. It is an expensive, dangerous metal hunk that sits almost 100% of the time in an unusable garage.

The car Suck

And self-driving them doesn’t solve almost any of these problems. That is Trouble. When we spend too much time focusing on the semi-mythical state of complete automation, we ignore the effective problems that lie ahead of us. Uber was successful because you can call a car by pressing a button. Despite the price, the lease is successful, as it is a separate control panel for the car. These are new transportation experiences.

So, where is the real opportunity?

I think companies like Zoox have interesting and compelling theses. Focusing on the morning experience, and designing a highly novel interface for telegidence, I think they have a real shot at delivering something useful out of a passion for self-driving cars. I think it’s important to understand, however, that their telegidence system is not a temporary bridge to get from here to there. The telegidence system and its supporting architecture are more defensible success for them than any algorithmic advantage. It combines with a model that provides an attractive vision that eliminates ownership. No … you know … just.

Don’t get distracted.

I haven’t used Zoox’s telegidence system. I don’t know for sure if they are more efficient than driving, but at least they are directed in the right direction. We need to stop thinking about self-driving cars as fully autonomous. While Level 5 autonomy is always around the corner, there is no need to think about all the awkward intermediate positions. The truth is that it is the whole point of awkward intermediate conditions.

This is the root of the problem of self-driving cars.

If you are an investor in the search for the first company to “solve” a self-driving car, you will be barking at the wrong tree. The winner is the company that can actually offer improved unit economics on vehicle operation. All closed track demos and all the vanity metrics in the world make no sense unless we solve that problem. We are dreaming about the end of the race when we don’t even understand how to fill the first step.

And obstruction Not at all Machine learning.

It’s user experience.

Slater is the founder and CTO of Victorof Indico Data.


Welcome to the VentureBeat community!

DataDecisionMakers is where experts, including tech people working on data, can share data-related insights and innovations.

If you would like to read about the latest ideas and latest information, best practices and the future of data and data tech, join us at DataDecisionMakers.

You might even consider contributing to your own article!

Read more from DataDecisionMakers

Similar Posts

Leave a Reply

Your email address will not be published.