“There Can Be Only One”: The Highlander Principle of Data Science
Despite (or perhaps because of) being one of the cheesiest movies I can remember from the 1980s, “Highlander” (1986) has had tremendous staying power in nerd culture. I attribute its enduring popularity to three things: a timeless battle between immortal warriors, a hard-hitting Queen soundtrack, and Sir Sean Connery’s inimitable one-liner: “In the end, there can only be a.”
It turns out that “There can only be one” has become a bit of a mantra in the tech world. For example, in 2009, a technology CEO wrote about innovation and the Highlander principle in Harvard Business Review. A few weeks ago, legendary tech journalist Kara Swisher discussed this at length with former Twitter CEO Dick Costolo on her “Sway” podcast. Speaking of the concept of a blockchain-enabled decentralized Twitter, “Any good tech nerd should know the Highlander benchmark,” Costolo joked. “If you don’t, you get kicked out of the club.”
A single enterprise-wide standard for AI development
As head of FICO’s data science organization, “There can only be one” has also been my mantra for a long time. This has been part of my discussion of model development for years, and in June 2021 I wrote about the Highlander Principle in a blog post about the importance of auditable AI:
Many companies suffer from many religions of data science – individual groups or, worse, renegade scientists marching to the beat of their own philosophical drum. In some cases, critical elements of the governance model are simply not addressed, which is worrying. Moving from research mode to production mode requires data scientists and businesses to have a firm standard in place. Since I…believe that innovation should be driven by the Principal Highlander (“There can only be one”), here are the questions your organization should ask to develop an auditable AI:
How is the analytical organization structured today?
How is the existing Analytics Leaders Governance Committee structured?
How is responsible AI being approached?
What is the status of the data ethics program and data use policies?
What are the AI development standards?
How does the company achieve ethical AI?
What is the company’s philosophy regarding AI research?
Is the company uniformly ethical with its AI?
(Above questions have been edited for brevity. For the full version, read my blog “Beyond Responsible AI: 8 Steps to Verifiable Artificial Intelligence.”)
Why “There Can Be Only One”
Many organizations, including data science teams, derail innovation with internal competition that fragments resources and energy. In a recent article on how to foster healthy intra-company rivalry, MIT Sloan Management Review has this as its first guiding principle:
1. Unify with a common goal. To engage in healthy competition within organizations, people must see themselves as united by a common purpose and a higher calling. At NASA, for example, employees’ deep belief that their work contributes to a higher purpose provides an effective counterbalance to a results-oriented and competitive internal culture. Every year for nearly a decade, NASA has ranked first in employee satisfaction among major federal agencies.
For data science organizations, the Highlander Principle not only establishes a single common goal – to create responsible AI that is innovative, ethical, explainable and verifiable – but a single detailed enterprise-wide framework on how to do it. That’s what AI governance and model governance are for: developing a singular business vision for fair and unbiased use of AI, and governing the path to get there with principles, processes and common tools. It’s not fast and it’s not easy. But if you want your company’s investment in AI to have the strength of “Highlander[SZ1] “, it’s true that “there can only be one”.
Follow me on LinkedIn and Twitter @ScottZoldi to follow my latest thoughts on AI and data science, and great movie picks.
Comments are closed.