I’m sure many of you are as excited as I am about all of the new innovations that are coming to Microsoft 365. We’re especially looking forwards to helping clients automate the metadata capture and content classification with SharePoint Syntex which I discussed in detail here:
Following its formal introduction at Microsoft’s Ignite 2020 conference, we are finally able to talk publicly about SharePoint Syntex...
By Rob Bath|2022-06-17T15:15:29+01:00September 23rd, 2020 |
SharePoint Syntex is the first of the products to be launched out of Project Cortex – I’m sure many will be keeping an eye out for the future release of the (currently unnamed) knowledge network and topic cards capabilities.
This future ‘knowledge network’ product is certainly going to be exciting many in the Knowledge Management community, however, my guess is that secretly many will be harbouring reservations. While AI offers huge potential (which I think everyone can see), I feel many fear it’s arrival might devalue or even negate the hard work that they put in.
This blog post is my attempt to put these fears to rest – and I hope to do so by making a subtle distinction: the ‘knowledge network’ will be AI-assisted, not AI-driven.
Together with AI
It’s best to think of AI as a tool that will intelligently help to undertake much of the heavy lifting and shifting – something that takes the strain and allows humans to focus on applying the finesse. This is especially the case when we are thinking about knowledge – as even the very best AI systems in the world still struggle to fully understand the value and intentions within our content. AI will be best at working at scale, finding relationships in large volumes of content – but it will still need human oversight to get the very best outcome.
What are AI's current limits?
Some of you might be familiar with the Winograd Schema – a test for machine intelligence (in a similar way to the Turing Test), which still provides a challenge for modern AI systems. The Winograd Schema is based upon the fundamental ambiguity of language – where humans can imply meaning that even modern AI struggles to understand.
Take the sentence:
“The firefighter carried the victim down the ladder, so that they could [survive/concentrate on] the fire”
Who is implied by ‘they’?
Notice how our understanding of whether ‘they’ refers to the firefighter or the victim is determined by whether the sentence includes the phrase ‘survive’ or ‘concentrate on’. A human can understand the meaning of ‘they’ without issue, but even advanced AI still struggles with comprehending this sort of implied ambiguity.
Project Cortex’s ‘knowledge network’ product will use Microsoft’s powerful AI capabilities to assess the content in your tenancy and identify patterns and relationships in the data. It will process large volumes of content that would take a human months or even years to evaluate. The patterns that Microsoft’s AI identifies will become the basis for the ‘Topics’ in your knowledge network.
What is a Topic?
Topics are a key aspect of Project Cortex’s ‘knowledge network’.
I find it easiest to think of Topics being a little bit like pages in Wikipedia – each topic is really a different subject that your organisation stores knowledge about.
As such, a topic could be a project, or a location, a product, committee, or team – effectively a topic can be almost any subject or theme and will vary between organisations depending upon the way you work.
The more that organisations assist the Project Cortex’s AI, the more value they will get from their Topics. Knowledge Managers will be essential part of this process – effectively acting as curators of their organisation’s ‘knowledge network’. Effectively, Knowledge Managers will be essential to add, extend and maintain all of the Topics in their tenancy.
What a knowledge network isn’t
Let’s tackle what I feel will be a common misconception – Project Cortex’s ‘knowledge network’ is not indented to be a substitute for getting the core information management principles right. If your files are largely unclassified and unstructured, I’d suggest you should be looking to fix this issue first, before looking to apply a knowledge network. Ignoring a document management mess and hoping for a future quick-win to tidy it up for you, will almost certainly see your organisation’s compliance, efficiency and collaboration continue to get worse until you address the core problem.
Instead, the way you should be looking a ‘knowledge network’ is more as a way of allowing your existing content to provide your organisation with additional value. It’s about helping to find and share information intuitively across your organisation.
Growing your knowledge network
A healthy ‘knowledge network’ won’t grow by itself – Knowledge Managers will need to work alongside AI to ensure that all of your topics work together to provide accurate and consistent knowledge.
There are a series of steps that I feel every Knowledge Manager who is working in conjunction with Project Cortex’s AI should really follow:
Complement the ‘knowledge network’, by appending additional topics that haven’t been identified via AI
Update topics with additional information and relationships. Add to the associated people and content – improve topic definitions
Remove superfluous topics – enable your staff to focus on the knowledge that is more important to your organisation
Establish stakeholders, governance and training to support champions across the organisation
Another thing I feel Knowledge Managers really need to think about is defining a lifecycle for their topics. It’s too easy to focus on the process of creating and fostering topics – but I’m fairly sure that pretty soon it will be clear that we need to consider ways to de-emphasise and depreciate topics.
Putting aside the (current) feasibility of the suggestion, but might we soon be discussing the need to archive a topic?
So where next?
While Project Cortex’s ‘knowledge network’ product hasn’t launched yet, it’s already very clear to me that we should be viewing it as a tool that we should work alongside. Knowledge Managers can let the AI undertake all of the heavy-work of finding topics and establishing relationships – but to get the most out of an AI-driven tool like this will require careful, planning, management and curation.
We should be looking to guide the AI, not let it steer us.
While we’ll have to wait a little longer for the release of Project Cortex’s ‘knowledge network’, there’s nothing to stop us getting ready for it first. How confident are you that your content is already well structured and consistently classified?
Intelogy can help you to improve collaboration and compliance by helping you to tidy and restructure your content. Want a solid foundation for Project Cortex? Why not speak to us today?
Having defined extensive Microsoft 365 EDRM systems and bespoke enterprise intranets, I specialise in overseeing cutting edge solutions that are tailored to meet customer needs. Providing leading expertise within the Information Management field, I enjoy helping organisations on their journey towards compliance.