Radical Reads

The Two Trillion Dollar Question – How to make data-driven infrastructure decisions

By Benjamin Schmidt, Ph.D., CEO & Co-Founder, RoadBotics (a Radical Ventures portfolio company)


Photo credit: AgileMapper, RoadBotics

Before U.S. Transportation Secretary Pete Buttigieg was tasked with the responsibility of over four million miles of American roads, he was responsible for the 550 mile road network of South Bend, Indiana. As mayor, before spending a dollar, Buttigieg oversaw the hiring of RoadBotics to support the management of South Bend’s infrastructure systems. The RoadBotics platform uses computer vision and mapping to survey road surfaces and enable, among other things, data-driven decisions to prioritize road repairs, saving governments significant tax dollars for one of their largest expenses.

Today, U.S. federal infrastructure spending is a hot topic due to a funding gap of more than $2 trillion expected by 2025. However, there remains little discussion on how to prioritize this spending. One of the biggest challenges is a lack of data on the infrastructure itself. From stop signs to sewer grates, critical infrastructure is everywhere. Keeping track of it is difficult enough. Gauging its need for repair? Even tougher. To help address this gap, earlier this month RoadBotics announced AgileMapper which uses AI to rapidly catalogue and evaluate municipalities’ existing infrastructure assets. With a better sense of what is at stake, decision makers can make data-driven decisions about where to spend their money.

By using everyday smartphones to deploy powerful computer vision and diagnostic tools, RoadBotics has assessed road networks for over 300 governments around the world, including 30,000 miles in North America. The goal is always the same: to provide cities big and small with the data they need to intelligently and objectively prioritize limited resources.

Crumbling infrastructure is not a construction issue. It is a maintenance issue. From roads and bridges to street signs and lights, infrastructure assets need to be constantly monitored and managed in order to continue functioning over long periods of time. AI-powered technologies can make the management of infrastructure systems more efficient and cost-effective, creating safer and longer-lasting infrastructure for years to come.

AI News This Week

  • The race to understand the thrilling, dangerous world of language AI  (MIT Technology Review)

    Natural language processing (NLP) is one of the most powerful technologies to emerge from recent advances in AI. For Cohere, a Radical Ventures portfolio company founded by former Google researchers, the responsible and safe development of NLP was a guiding principle behind its creation. As this article points out, Cohere is bringing the most sophisticated NLP systems “to any business that wants one—with a single line of code.” To deliver on this ambition, Cohere is investing in both technology and teams to ensure the models it deploys for customers are safe and accountable. The company is also creating an advisory council of external experts to help it create policies on the permissible use of its technologies.

  • Under the AI hood: A view from RSA Conference   (VentureBeat)

    While realizing the full potential of AI in security requires long-term investment and collaboration between highly-skilled security and machine learning talent, this year’s RSA Conference featured an entire track dedicated to security-focused AI. AI is an essential technology within the cybersecurity sector, enriching analysts’ abilities to review mass quantities of data, prioritize alerts, and make basic security decisions at scale

  • These Ex-Journalists Are Using AI to Catch Online Defamation  (Wired)

    Global cutbacks in journalism have left newsrooms without the resources to check the spread of disinformation. Applying AI technology to this issue may result in “spell-check” for disinformation, as well as legal liability for harmful and illegal content such as defamation. This would be an invaluable tool for both journalists and readers.

  • The Race for AI Supremacy: U.S. vs. China   (Pairagraph)

    Economist Carl Benedikt Frey and Professor Thomas H. Davenport debate whether the U.S. or China will lead the world in AI research and applications. The dynamic discussion underlines that both countries are allocating billions of dollars toward the development of AI as a crucial technology for future prosperity.

  • Weird dreams train us for the unexpected, says new theory  (The Guardian)

    Do humans dream to avoid “overfitting”? Typically, neuroscience acts as a muse for the design of AI architectures. In this case, a neuroscientist’s theory for why humans dream was inspired by training AI models to avoid becoming too familiar with the training data. In short, a bit of randomness might be good for us.

Radical Reads is edited by Leah Morris (Senior Director, Velocity Program, Radical Ventures).