Seven lessons learned about knowledge management
This is the sixth and final post in our guest blog series by Esteban Kolsky of thinkJar. See the original post for an introduction and a list of all posts in the series.
Throughout this project, and the many interviews and conversations about the topic, we came across some phenomenal experiences. The lessons we learned while interviewing practitioners and discussing the concepts were simple but very effective. With regard to using KM aided by AI, use the following seven lessons to frame your initiatives:
- AI is about data, KM is about knowledge. Neither one works independently, and neither can replace the other. You must continue your KM efforts, augmented by AI, to manage content and you must learn to use AI to manage data and the variables that affect the usage of that content in context – which is what knowledge is.
- Takes time to do it right. Don’t be fooled by the hype and the many vendors promises of “quick results”. While you may be able to solve a simple problem by adding AI to it, it will be short lived. Just like with KM, proper initiatives with long-term planning and maintenance (or training in the case of AI) are necessary before results become sticky.
- There is no one-solution-solves-all-your-problems approach. This is one the largest gaps between technology promises and reality. While one out-of-the-box solution may be easy to deploy under certain controlled conditions, the use of AI to augment KM requires extensive planning, training and testing.
- Know your outcomes. The success of AI is measured by ensuring that results are better than they were in expected outcomes. There are three potential outcomes for AI: optimization, personalization, and automation; they are different and vary in complexity and timing to get right. Know which outcome you are going after, and plan accordingly.
- Don’t be fooled by the past. You may have done many AI-like or AI-related things in the past: natural-language processing, fuzzy logic, decision systems, or others. The new AI, within the framework introduced in section one above, requires using the lessons learned while doing that, and extending it to a full-enterprise, end-to-end leveraged model to deploy AI to specifically augment KM. Don’t forget what you learned.
- Lather, rinse, repeat. It bears repeating: the results intended and expected the first time an AI initiative is launched are unlikely to either be obtained or be the main reason for adopting AI. It takes several iterations to align the understanding of the tools, the definition of the needs and outcomes, and the testing of many models to find the combination that works best. Iterate.
- Maintenance is, still, the most important part. As one of the practitioners we interviewed said: make sure they understand that launching is nothing compared with ensuring it works well, it continues to learn, and it does the job intended. What he said.
There are, I am sure, many more lessons that are either critical or that have been learned the hard way. While we may have had AI around for close to 40 years, we are still learning what it can (and cannot) and what it should do. We have been working with KM since the 1960s and are just starting to figure out a more efficient model for using it in the enterprise.
Patience is the most important lesson learned.