The AI Governance Tool
A simply boring attempt at making AI governance and risk management accessible
If you’re infantry in the Singapore army, you would probably have encountered this protocol to shout “gap gap gap” when you breach a fortification.
I always thought it was kind of stupid. Since shouting it was asking for the bullets to be aimed at you. I always felt I was lucky to move from infantry to be a scout. Scouts never ran towards bullets.
But I think I understand the need for it more now. Shouting it in the army helps make people aware of the gap that you are moving towards.
In one of my past weekly reflections on LinkedIn, I said three gaps kept surfacing. The common language gap. The contextualisation gap. The last-mile gap. Same need for awareness.
Frameworks exist. Guidelines exist. I wrote some of them. But translating frameworks into something relevant for one’s context is not easy. Especially for individuals and smaller firms.
Last week, MAS launched the MindForge AI Risk Management Toolkit. I was deeply involved as it needed to be aligned to the MAS AI Risk Management Guidelines I wrote.
The MindForge AI Risk Management Toolkit is an operationalisation handbook with case studies. It’s an attempt at bridging the messy middle I keep writing about. But I suspect, aside from large financial institutions, it may still be a lot to chew on for everyone else.
Last week, a friend who leads manpower development also asked me something similar: what about everyone else?
This is my early attempt at an answer.
GOT, not Game of Thrones, but an AI Governance Tool.
Describe what you are using AI for. Get a tailored governance pack in minutes. The standards that apply to your situation. The controls to put in place. How to implement them. What evidence to prepare.
Built on public information. The MAS AI Risk Management Guidelines I wrote. The MindForge toolkit that just launched. The US Financial Services AI Risk Management Framework. The EU AI Act. NIST AI RMF. ISO 42001. And other international frameworks. Cross-referenced and connected. And a little of my own perspectives.
The AIRG tells you what to do. MindForge shows how the industry is doing it. GOT asks what you specifically are doing, and tells you what applies to you.
Not just for the large institution with a full second line (and third line) of defence. For everyone else. Individuals. Small firms.
It’s an early prototype. It will almost certainly have errors. And a little slow - around 45-60 seconds for each report. And I think it’s still a little iffy for individuals.
But I tried to ground it in real frameworks and design it for real situations so anyone, not just banks.
Please try it. Tell me what’s missing. Tell me what’s wrong.
#Mindforge #NIST #AIRiskManagement #SimplyBoringAI #Grounding #AIRG