AI: The Good, The Bad, The Ugly with Janet Xiu

5/11/26 @ 1.55 pm GMT

What boards are being asked to approve, and what to ask before they do.

You cannot govern what you have never evaluated. Most boards are shown that AI has been adopted and is being used. Very few are ever shown evidence that it works, with proven return on investment.

The Good. Where AI is genuinely paying for itself, and why it is rarely the use case that makes the press release.

The Bad. What happens when you actually test the AI a company has bought. The demonstration always works, the deployment often does not.

The Ugly. Shadow AI, company data leaving the business through free tools, and AI agents being given real permissions in real systems.

                       Why this reaches UK boards. The EU AI Act applies to companies with no EU entity at all, where the output of their AI system is used in the EU. Its transparency obligations came into force on 2 August 2026 and were not postponed.

What a chair can do. Four questions to put to the executive team at the next board meeting, and what a good answer sounds like.

Bio

Janet Xiu Shi is an AI quality, evaluation and governance advisor based in Ireland. She works with boards and leadership teams on the question most AI conversations skip: does this actually work, and what evidence should we be asking for. She is a judge for the AI Awards Ireland 2026, an Industry Expert Lecturer in Cloud, GenAI and Data at IADT in Dublin, a national instructor for Ireland's state funded QQI (Quality and Qualifications Ireland) artificial intelligence programmes, and a fractional Strategic AI and Cloud Advisor with Connectd. She built and hosts Maya, a live AI agent, in the EU to meet the AI Act's transparency obligations.

janetxiushi.me | maya.janetxiushi.me | linkedin.com/in/janetxiushi