An executive at Google Cloud is warning that the transport industry isn’t getting the best out of its data, because it doesn’t have the expertise to do so.
Strategic Business Executive Chris Haas made his comments in an interview with the Highways Voices podcast about the AI technology that that allows software applications to become more accurate at predicting outcomes without being explicitly programmed to do so. The interview will be featured in a podcast published on Wednesday 8 June on the Highways News website.
Talking to Highways Voices presenter Paul Hutton at the ITS European Congress in Toulouse, Mr Haas commented that transport has not “had the talent influx of the data scientists and people that would be needed.”
“The actual roadway operators themselves don’t have the capabilities or the skills to really do machine learning,” he said, before warning that the technology is not the silver bullet. “Everyone thinks machine learning will fix their problems, regardless of what their data looks like. So they’ll throw spaghetti data at it and expect it to give you some really amazing insight. That’s just not the way it works. It starts with organising, understanding your data before you can apply machine learning.”
In the interview Mr Haas talks about the risk of data bias, where data is used to try to prove an expected outcome, rather than leading the thinking, and the need for consideration to be given to the fact many of the systems the industry uses are using old technology.
He also concentrates on best use of data. “Before I even start thinking about how am I going to use the data and help me solve my business needs?” the former developer comments. “Does anyone say business here I think of like a road operator public authority that you know, their business, what are they delivering for citizens or for drivers? So I reverse that, and I call it a use case. I think about what’s the thing that we’re delivering, what’s the business need? And then I figure out what the technology is. So just reversing, I think, our traditional way of thinking about applying technology to problems, which then helps you really think of the end result first, then you figure out what data you need. And then you start analysing the data to to help you deliver on what that outcome is.”
Highways Voices will be published at 9.30am Wednesday 8 June. You can ensure you never miss an episode by subscribing to Highways Voices free on Apple Podcasts, Spotify, Amazon Music, Google Podcasts or Pocket Casts.
(Picture – Yay Images)