Building Ethical AI: Reid Blackman on Automation Bias, Deepfakes, and Government Priorities

Quotes From Reid Blackman on AI and Ethics

“Think what your organization’s ethical, reputational, regulatory and legal nightmares are. Define those ethical nightmares in a comprehensive way, and then systematically and comprehensively put in controls to stop them from happening.”

“It’s usually the Chief Data Officer, Chief Analytics Officer, Chief Information Officer,  Chief Technology Officer, it’s usually someone on the tech side of the house. But then I always have to tell them “you guys are spearheading this, but you also need to bring in people from risk, compliance, legal, cyber security and HR. At the end of the day it’s a cross-functional effort, an enterprise-wide effort. You have to have cross-functional senior level buy-in.”

“There are lots of statements out there by lots of organizations, governments, non-profit, corporations that say, “Here’s our ethical standards,” and they’re strikingly similar to each other. “We’re for fairness, we’re for transparency, we’re for accountability, we’re for non-discrimination, we’re for respecting privacy.” They’re keywords.”

“I think one way to specify what those (AI) guardrails are is to first start out with specifying what the nightmares are.”

“Some people say “oh AI ethics, it’s about social benefit and positive social impact. That’s great, go do that, but that’s not priority number one. Priority number one is “let’s not blow everything up.”

“The thing that businesses need to understand is you don’t have to cater to your bottom line, this is not about stifling innovation, it’s about stifling really ethically bad innovation.”

“We already have a human rights framework that’s internationally agreed upon. It’s required of all organizations to do a human rights assessment for all the AI that they’re creating throughout the AI lifecycle and to engage in robust risk mitigation, human rights violation mitigation strategies and tactics throughout that, and to document it and to be transparent about it with the relevant authorities.”

“It’s about an ethical risk appetite that’s compatible with business risk appetite and operational risk appetite.”

“For government regulators, at a minimum, their function is to protect people from the worst. Let’s put in regulations and laws to make sure that people are protected from the worst.”

“The EU is making a great stride in passing what’s called the EU AI Act, the European Union Artificial Intelligence Act, which is the first set of regulations around AI.”

“The Continental philosophers can be more literary than analytical. And I think that the analytic approach to philosophy, which is the dominant approach in the top research institutions, lends itself just to the quick grasping of lots of concepts and how they relate to each other.”