Hello, welcome to the 8th issue of the Machine Learning Assurance Newsletter, a newsletter at the intersection of machine learning, regulation, and risk. I'm Anthony Habayeb, the co-founder and CEO of Monitaur.
In this issue, I find it quite interesting – and somewhat concerning – to see this separation between accelerated regulatory interest and guidance around responsible, auditable, objective oversight of AI and ML and then VentureBeat's article which references a survey that found 65% of executives can't explain how their AI models work.
What fits in the middle?
We at Monitaur believe that in between you can build practical governance and enablement of assurances. But it's complicated, and as you read through these articles, you can see why there's so much angst right now in the broader AI market talking about fairness and bias. There is such a gap between regulatory interest in protecting consumers, what they would like to see from AI, and then the operator's readiness and ability to effectively communicate about what their technology is doing, even internally to have confidence that they can explain how their AI is making decisions.
It's not impossible. You can absolutely do things today to start to build better controls. Read through these articles and think about how you are, as a company, are implementing good oversight, controls, and governance across your systems, as well as how that effort translates into the signals that regulators are seeing or thinking about the use of this technology.
Hope everyone is having a great day, and we'll talk to you again soon.