Interested in a ServiceNow event built for developers? Registration for now[dev]26 is officially open!

The Real Talk: Your Knowledge base will never be perfect

dereknohr
Tera Contributor

One thing I keep hearing as I talk to my peers about turning on AI is:   I'm not sure that our knowledge is in a state where it will make AI beneficial. 

 

I get that but I think it's more of a delay tactic than an actual reason to not get started. 

 

Yes, KBs exist from systems which were retired years ago

Yes, the process changes so frequently that the moment you hit publish, it needs to be edited again

Yes, there are whole businesses processes not in/connected to ServiceNow

 

Despite your view of your data, is it enough to not take the first step?   

 

Turning it on and prompting the common user questions will tell you if your data is garbage or if it's  performing 'good enough'.   Don't let perfection get in the way of you getting started.

 

What's the worst which would happen if you go live with sub par knowledge?   Scare me straight.

 

 

9 REPLIES 9

briannuernb
Tera Contributor

I would start with a smaller group with access prior to turning it on for the masses.  This is how we approached it to ensure much of our knowledge was ready. Think about your top questions that your service desk or your users are asking/searching for.  Test it out to see if the answers it provides make sense. If the answers are off the wall, revise the KB.  Any AI is only going to be as good as the foundation it has to build its information from, and nowassist is double dependent on that foundation. As far as scaring you straight, it depends on the industry you are in 🙂

exactly...I limited to just IT...you know the folks who 'get it'

Nia McCash
Mega Sage

One of the worst things I can think of is that the AI implementation ends up being garbage (eg. a chatbot performs poorly and gives a useful answer only half the time at best) and therefore dilutes confidence in any future AI implementations. 

But still, I take your point and the reminder that we shouldn't let perfect be the enemy of good.

this is something I struggled with:  AI in the workplace is NOT the same as AI in the public.   there needs to be lots of reinforcement to the employees that there is a difference.  Even so....if it goes REALLY badly...you do destroy trust