AI reliability changes how you watch your systems

In this clip from an AI roundtable with Gremlin, Nobl9, and PagerDuty, Mandi Walls talks about how AI shifts how you watch your systems to keep them reliable. Find out how to improve AI reliability with Gremlin → https://www.gremlin.com/solutions/improve-ai-reliability

FULL TRANSCRIPT:
With any system, your best laid plans, once they meet the users, might explode. Right? So what we also see is the opportunity for checking the system's answers if what the users are doing changes. So release something out into the world and the users start doing something slightly different with it that we didn't necessarily train the model on or the stream of incoming data that we're sending to it starts to change because of user behaviors or whatever.

That all has an impact, too. So part of machine learning observability and watching all these things over time is watching not only the outputs, but also the shape of the inputs and making sure that the model that we built is still fit for purpose for the data as it changes. And I think that's maybe a new challenge for a lot of folks too.