News & Views

Interview / Accelerating AI

by Contagious Team

This story originally appeared on Contagious I/O, our customisable research platform featuring the world’s most innovative, creative and effective campaigns and marketing ideas

Auto brand establishes research centre to advance intelligent vehicle technology – we speak to the team behind the project

Toyota has partnered with MIT and Stanford University to help use artificial intelligence to make driving smarter.

The $50m investment from the brand will focus on how computer science and human-machine interaction can reduce injuries and fatalities on our roads. The joint research centres will be located on both university campuses, and will focus on assisting drivers rather than automating the entire process (i.e. driverless cars).

The Contagious I/O team caught up with Steve Eglash, executive director of Stanford’s Artificial Intelligence Lab, to learn more about how the partnership intends to create meaningful change in the auto sector.

What’s the purpose of this partnership?

It’s about creating a future for how humans will interact with intelligent machines, like automobiles or robots. We believe that, for this to be effective, it’s got to be human-centred and the machines have to be context-aware.

So in the case of automobiles, for some years now we’ve heard about autonomous vehicles. But the most thoughtful players in the space believe that the more interesting and important application for the next couple of decades is probably assisted driving in complex environments. There are many complex environments that are important but also challenging for intelligent automobiles.

Can you give me an example?

Imagine a situation where the road is covered by snow and you can’t see any markings or the edge of the road. You or I can pilot a car safely down that road, but machines still can’t do that. Or imagine seeing an object in front of you when you’re driving at 100km/h and you have to decide whether it’s a blanket, that you’re safe to drive over, or a rock that you must swerve to avoid. Cars struggle to make decisions like this in a fraction of a second.

Beyond looking at the world around you, you’d also like the car to look inside, to understand whether the driver is engaged in driving, or distracted, inebriated or sleepy, and then intervene in the appropriate way.

When will this kind of technology start to become commonplace?

You need to view all of this as a gradual and progressive transition. We’ve already got cars that do a pretty good job of autonomous driving in certain situations  like driving down a motorway where conditions are good. We have things like adaptive cruise control, and cars that keep themselves in their lane. The vision that many of the most sophisticated companies share is a vision that we’re simply going to see an increased penetration over time of both assisted and autonomous technologies  let’s just call them intelligent vehicles.

What are the biggest obstacles facing you and your team?

The biggest challenge of course is intelligence. We need hardware and software systems that can identify cars, pedestrians, and bicyclists, and then predict their behaviours with high accuracy. We need cars that can make decisions in the presence of uncertainty and then make adjustments and corrections in real time. If we do this well, then our cars will become much more than tools; they will become our assistants and extensions of ourselves.

This future relies on sensors  everything from video cameras to sensors that use radar technology to figure out what objects are around you and predict their behaviour  those sensors are going to have to get less expensive for the technologies to penetrate automobiles entirely.

Another challenge is privacy and security. Your car is among one of the most connected things you own. Thanks to your smartphone, your car even has the ability to know what you’re doing when you’re not driving. It knows when you go into the pub for a beer, or a jewellery store to buy an engagement ring for your fiancé. So what should your car do with that info, and how can we protect people’s privacy? Of course, we’ve all heard of cars being hacked into, so how can we make sure that there’s good security?

Car companies are incredibly concerned about this because they know that if they get it wrong, if one car brand screws up and develops a reputation for spying on its owners, then people are just going to flock to other brands.

On the other hand, if carmakers can get it right, if they can design cars that we love as much as we love our smartphones, then they know that people will flock to their brand. So figuring out how to make technology that people love, rather than perceive as spooky, is another big challenge. All of us in technology and in society have a responsibility to engineer the future in such a way that people are glad to have these technologies, not the other way round.

What does this evolution mean for the auto manufacturers?

The future of competitive advantage in the auto industry is no longer going to be bending sheet metal and making drivetrains, the way it has been for the last 100 years. If that’s the only thing you’re good at as a car company then you’re going to be stuck at the low margin-end of the food chain. So all car manufacturers have figured out that the path to being profitable in the future increasingly relies on all the things other than the sheet metal and the drivetrain. And chief among these is intelligence.

Why should consumers want more intelligent cars?

It’s important on a number of levels. Safety  the number of people killed in traffic accidents is astonishing. The only reason we tolerate it is because it’s always been this way. But with the help of these technologies, a day will come when people will look back on the current period of time as the dark ages. They’ll be as shocked that we tolerated traffic accidents as we are today when we look back on the medical practice of 75 years ago, before people appreciated the need of being clean and sanitary when doing surgery.

An additional benefit will be to provide mobility and transportation to people who currently can only take advantage of it through things like public transport, taxis and ridesharing services like Uber. This could be the elderly, people with injuries, younger people. So as cars get more capable, drivers won’t need as much capacity to safely hop in them. It’s just a small step further to imagine different kinds of robots or vehicles that can deliver packages, for example.

Beyond that, we can move on to benefits that make life easier, more efficient and more entertaining. Imagine sitting in traffic in London and not having to concentrate on the road ahead.

These are the sort of things that, when we have them, it’ll be hard to realise life without them.