BlueConduit’s co-founder Jacob Abernethy, a Research Scientist at Google Atlanta and Associate Professor in Computer Science at Georgia Tech, recently co-authored an article in Fortune Magazine titled, “Your AI products’ values and behaviors could make or break your business. Here’s what you need to do to get them right.”
In the article, Abernethy and his co-authors argue that building values-based decisions and behaviors into AI tools “will increasingly become a source of differentiation and competitive advantage” and they share guidance for companies on how to get AI values right.
For Abernethy, this is not an academic argument. Since co-founding BlueConduit in 2019, he’s worked closely with our internal teams and external partners to embed values-based decisions and behaviors into BlueConduit’s technology tools, including:
For Abernethy, this is not an academic argument. Since co-founding BlueConduit in 2019, he’s worked closely with our internal teams and external partners to embed values-based decisions and behaviors into BlueConduit’s technology tools, including:
1. Industry-Recognized Best Practices for LSL Inventories
In the Fortune article, Abernethy and co-authors tell us to “define your values, write them into the program—and make sure your partners share them too.” The BlueConduit team, alongside our partners at ASDWA, published the white paper that defines the industry best practices for using predictive modeling to develop Lead Service Line Inventories. This approach, embraced by the EPA in LCR and LCRI guidance, is based on best practices in data science alongside our direct experience with LSLI and replacement.
2. Equitable, Ethical Data Practices
The article also states that “[i]dentifying the right values is a dynamic and complex process that must also respond to evolving regulation across jurisdictions.” BlueConduit’s technology tools, supported by a team of expert Data Scientists, are built on a foundation of good data management practices, compliance requirements, representative sampling, high quality statistical models, and rigorous testing to ensure we identify and address any potential bias in customer data.
3. Act as a Conduit for Communities
Abernethy and his co-authors further note that “alignment with values requires the further step of bringing partners along.” Communities are unique and there is no one-size-fits-all predictive model that can generate high accuracy results for every community. At BlueConduit, all of our technology tools and processes are supported by our expert Data Science team using a human-in-the-loop model to ensure high accuracy, highly local models that reflect unique customer preferences and needs.
4. Transparency in Data Practices
The Fortune piece emphasizes that “maintaining an AI product’s values, including addressing biases, requires extensive human feedback.” Predictive models rely on a series of data decisions and choices. At BlueConduit, we work hard to ensure our modeling process is not a black box; customers receive Data Summary and Data Protocol reports, among others, that outline specific decisions and methods used to generate their unique model.
At BlueConduit, we’re proud to follow Jake’s leadership by clearly defining our values and ensuring those values are built into our machine learning tools as well as our broader company approach.
Interested to learn more about BlueConduit’s predictive modeling tools for lead pipe identification and replacement? Get in touch!
