We have enabled more tags. When you post a discussion or reply to a question, add tags. They have full search capability. Try it!
Listen to the main takeaways from our latest TSIA World Envision conference at TECHtonic podcast now!

Learnings based on AI / ML

Do any of you have a working model of training recommendations based on AI / ML recommendations / personas / engineer learning behaviours in automated way ?


  • Alexander Ziegler
    Alexander Ziegler Founding Member | Expert ✭✭✭

    Hi, the easiest way to start recommendations is to use a "You have taken course X - and most other people who have taken course X afterwards took course Y". Most Learning Systems are able to collect this data, and you can relatively simple program this. It is not perfect - but a good first step. Next would be "Users who have taken X, Y and now Z usually also take course ZZ" - which is quite strong, as it kind of removes the need to program roadmaps, once you have enough data. Those approaches are called "Weak AI" - and the ideal starting point.

    Afterwards, you could implement profiles of users, and then give recommendations based on roadmaps, profiles and other users. But implementing step 2 from above gives already quite good recommendations if you have enough data.

  • Thanks Alexander , any vendors who do something like below ? -

    1. Look at engineers case/SR data
    2. see how the engineer had handled the case
    3. based on the way case was handled , suggest trainings - could be professional / technical / soft skills
  • Alexander Ziegler
    Alexander Ziegler Founding Member | Expert ✭✭✭

    Hi, I have doubts that this exists - and my recommendation would be to do an analysis of NPS+NPS Comments (if you do not use NPS or without comments then introduce NPS and/or comments) - and have a manager taking an NPS report of the lowest performers and look to the comments. This is something simple and powerful and easy to implement - and much more powerful than trying to go with AI against the case data.

  • Thanks Alexander , we do have NPS and CXSAT along with verbatims but they are manual process to determine gaps and manual process of training identification.

  • Lawrence Kaye
    Lawrence Kaye Member | Guru ✭✭✭✭✭

    Hi, to support Alex's statement, I just finished a report recently (coming out before conference) on the use of Metaverse in training and, with all of the discussed cutting edge technologies, I didn't see the application of AI to drive recommended training to the level you're looking for. I did encounter one example of bots being used to answer training FAQ to cut down on operations expenses.

    In other conversations with Education Services members, Maria and I have heard references to one platform, Degreed (www.degreed.com). It may be something you could explore.

  • John Ragsdale
    John Ragsdale Member | Guru ✭✭✭✭✭

    My 2 cents from the technology standpoint: What you are describing is supposedly at the heart of the new Learning Experience Platforms (LXP), though this remains an emerging technology area. Our partner Thought Industries has recently introduced a new LXP to join their suite of learning management and content development tools, so may be worth checking out. https://www.thoughtindustries.com/

    One other thought. Our partner TheLoops does correlation analysis using a virtual data lake pulling information from multiple systems. As an example, they are telling tech companies which features used by customers in a product correlate to value realization, renewals, and expand selling success, so they can prioritize those features in onboarding. They could easily analyze learning content consumption patterns and correlate to adoption, renewals, etc. https://theloops.io/

    Both of these partners will be exhibiting at our TSIA World: Envision show in Las Vegas coming up in October, so if you are attending, be sure to check them out!

  • Hi Lawrance , we already use degreed but it does not handle AI/ML part.

  • Sure John , will be there and meet these teams for sure.

  • Lawrence Kaye
    Lawrence Kaye Member | Guru ✭✭✭✭✭

    Thanks for the data about Degreed - I'll make note of that for the future.