How machine learning could impact society
Machine learning has been around for quite a while. But are you actually taking notice of it?
A large number of systems run on this application of artificial intelligence, which basically, as defined by Forbes, is giving machines access to data and let them learn from themselves.
“Machine learning is a huge game-changer,” Patrick van der Smagt, director for artificial intelligence at German powerhouse Volkswagen, told Khaleej Times on the final day of Artelligence 2018 in Dubai on Thursday.
“It’s been there for about 30 years, but it’s only now that it’s advancing because of computing power and renewed interest.” He says the challenge regarding this area is that there’s “no big possible feasible impact at this moment”.
Van der Smagt, who also gave the keynote address, cited the example of voice systems found in smartphones; people don’t usually use them because the impact “isn’t very clear in such a way”.
“Societal impact don’t dome abruptly,” he says, “but people gradually get used to it as they become available and people won’t notice it immediately, just like the smartphone; who would’ve thought 10 years ago that it would become a commodity today?”
Van der Smagt pointed out that machine learning brings huge impact to several industries. In the automobile sector, it can do much more complex things such as controlling autonomous cars in simulated environments based on driver experience.
He cited a specific example in the Netherlands, where the country had integrated transport strategies with respect to timetables — even informing passengers if certain train cabins are full so they can occupy those that have available seats. Developing these systems requires strong infrastructure — and machine learning is making complex tasks easier.
“In the past, anytime when you want to solve a problem, or control or influence a system, what you did is that you wrote down an
There’s a clear mindset in saying that ‘let’s not make sure that everybody has a car, but let’s make sure that everybody has the best possible way to get from point A to point B given the time and situation’ Patrick van der Smagt, Director for AI at Volkswagen
equation or run a simulation of that system based on engineering knowledge,” Van der Smagt said.
“That works nicely, but it only works nicely when the system is simple enough to do that and when the engineer is smart enough… it can be very expensive it doesn’t scale up.”
Today, what we can see is that modelling of a complex system is done through machine learning, to make predictions based on data rather than based on engineers, he added.
And he has an interesting insight on how this can also lead to a revolution in the auto industry.
“There is a mindset moving away from building cars to providing mobility services,” Van der Smagt says. “Of course companies are still building cars, but the distinct plan is, maybe 10, 20 or even 50 years from now, there will be a point where not everybody will have one or two cars anymore because you don’t need or want it, because maybe cities will be full [of vehicles].”
“There’s a clear mindset in saying that ‘let’s not make sure that everybody has a car, but let’s make sure that everybody has the best possible way to get from point A to point B given the time and situation’,” he stresses.
He added that, if achieved, it would be a “much more sustainable service” when you combine personal cars and autonomous cars with public transport and bicycles, and the whole infrastructure.
“Put that in a package that adapts to individualised needs and deliver that as a service. I believe that’s where we should go in the coming years.”