Can AI fix Knowledge Flows?
Rebecka Isaksson
KM Expert | Keynote Speaker | Podcast host??| Microsoft MVP (Microsoft 365 Apps & Services)
This is the final chapter in my mini-series on Generating Business Value by unlocking Knowledge Flows – what my independent business KnowFlow Value specialises in.
Subscribe to my newsletter, for automatic updates, here: KM Strategy put to work.
Debated by many Knowledge Management and AI experts is whether AI can in fact solve Knowledge [Management] problems. I would argue that it cannot.
Not unless we apply it with the basics of Knowledge Management as the foundation. But combine any Generative or other AI service, with KM fundamentals, it will have a much bigger impact on revenue streams, employee engagement and customer satisfaction. Without KM, it will be another productivity enhancer at best.
And talk about lost ROI, if AI is simply reduced to another productivity tool.
Let’s take a look at AI’s impact on Knowledge Flows, through the lens of Knowledge Management, and the four most commonly referenced types of knowledge:
领英推è
- Explicit Knowledge – knowledge that is captured in documents, images etc., thus converted into information.
- Tacit Knowledge – the knowledge we gain from practicing something over time, e.g., experience, which can be transferred.
- Conversational Knowledge – a great knowledge sharing mechanism as it allows us to enhance knowledge by learning from others. This can to some extent be captured but has limitations as soon as it is.
- Implicit Knowledge – knowledge that cannot be captured or transferred, simply because we aren’t even aware that it is there or why, e.g., intuition or emotions.?
There is no argument that AI can do wonders for Information Management. Not only can it help us organise and structure documents, and the information within them, but it can also help us interpret semantics and convey meanings of the information to us. It can even create information and documentation, by aggregating massive volumes of data, at scale.
In terms of experience, AI can absolutely be a great help in transferring, or broadcasting, what we have learned to others. However, we need to keep in mind that a broadcast (or recording) is not dissimilar to written down knowledge, as it is ultimately a form of one-way communication. It has advantages over “documents†in the sense that it incorporates implicit aspects of knowledge, like tone of voice or facial expressions though and we know there are AI services that can already interpret those too.
Chats and group discussions, or conversational knowledge sharing and exchanges, can be analysed and summarised by AI, which is huge progress, but still has limitations because it translates conversational knowledge into information. The value diminishes as the information then becomes subject to one person’s interpretation, making it similar to a user guide or instruction manual.
When looking at AI applications to extract information or knowledge from the implicit side of knowledge, we are at the verge of a potentially very steep and slippery slope. I would urge you to proceed with such AI applications and services with great caution and especially consider who owns the knowledge, once it has been interpreted or codified.
In the end, it boils down to this: do you still own your [implicit] knowledge, or does the corporation, which hosts the service now own it?
This is one of the main reasons for the EU’s AI Act and similar guardrails are being put in place by other nations and institutions. I am a firm believer in data solidarity, over data integrity, within areas like medical research or crime prevention, but that does not necessarily mean I think we should even try to codify implicit knowledge, and I hope this piece will spark some conversations here.
I am very curious to hear what my readers thinking is on this hot and incredibly important topic.
<<< Subscribe to my newsletter, KM Strategy put to work, for more articles on AI & KM. >>>