Bryan’s Blog

Testing Inputs as Well as Outputs

A quick question: When you review a proposal, analysis, or AI-generated output, what are you actually testing? Most people focus on the answer. Does it look right? Is it well expressed? Does it align with what we expected? All reasonable checks, but they miss something more fundamental, which is the answer is only as good

AI Is Exposing an Underlying Problem

“What are you up to with AI?” I’ve been asking that question a lot lately, and most responses focus on use cases, productivity gains, which tools to prioritise and, of course, AI governance. All are valid and important conversations BUT perhaps you need to be having a different conversation. Many organisations are treating AI itself

Self-Disruption is the New Normal

I recently asked myself: “Am I surprised at the completely different speeds AI is being adopted today?” And my answer did not come easily. I landed on “No”. For context, I was not surprised a year ago about the different speeds. However, as of today I continue to come across organisations that are not adopting

Loopy Humans in the Loop

Can the humans in the loop be trusted? How often are you hearing lately “We’ll always keep a human in the loop”? This is recognition that AI can’t be blindly trusted, even when it is a narrow use case, as your AI model can still change if the operating environment changes. But can we trust

Risk Assessing AI

While many of the risks associated with AI are now better understood, some organisations continue to experience undesirable outcomes with their integration. Take this little doozy involving Microsoft 365’s Copilot, which resulted in widespread breach of users’ privacy protections. Which brings me to ask, ‘How are you assessing the risk associated with the implementation of