Do current Information Management standards serve AI tools?
This post is a response to thread created here by Tom Grey on Linkedin (Feb 2026) - https://www.linkedin.com/posts/tomgray86_are-our-standards-ready-for-ai-its-not-activity-7426591967528189952-uM4v?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAGaJ-YBOdTiifusZece7fIMAWbCJkHXiJg
This is a timely post and I want to offer some reflections that hopefully align with this thinking.
I generally agree that standards are variable - both in their need and the contexts in which they are applied. And I think this is the crux of it - AI, both in its application and its ability to pull in data from across any domain seems to be immediately challenged by this truth.
How we think about standards and their implementation is (in my opinion) still pre-AI in the traditions and conventions being applied. Generally such thinking tends to immediately focus on particular internal priorities that are unique to the environment that has called for them. So, when you look at it organisationally, a group of organisations in one sector will have standards that relate to that particular sector (no surprise).
What’s interesting about point 2 is that when you look at different sectors (say if you contrast Defence and security with Industrial Manufacturing for example) they will both have different standards for different things. Obviously some of these are unique to the sector, but many are not - especially with AI and with its less popular cousin, Information management.
Because AI is such a leveller - are we now seeing the need to try to find a general consistent approach to standards and their utilisation? This is obviously a big thing.
To help with our thinking today and potential options to help us think better about AI - I generally suggest applying a framework that models all relevant enterprise activities and incorporates as many standards as possible before the enterprise applies its own preferred, often sector-specific, process or standard. This helps us think better about AI and our potential options (and even Information Management).
Methodologies such as this https://github.com/Apollo-Protocol/information-requirement-methodology (that start agnostically to define the best approach) - are extremely helpful here and they (hopefully) stop us ‘baking in’ a conventional standard that can cause a large amount of trouble down the line for data and information management (especially using AI tools)
As with some many things ‘model based’ thinking can really help here, but I fear the challenge is starting it early enough and at a high enough level to get a proper understanding of all the activities your enterprise undertakes (all of which are currently being shaken up across the board due to the impact of Gen AI).