Engaging Analytics & Humility to Mitigate UncertaintyAdd bookmark
Machine learning models learn from historical data and use the patterns they discover to make predictions about new data. In an ideal world where conditions are always consistent, it would be easy to have a great deal of confidence in every single prediction. But as we all know, especially today, consistency and stability are luxuries that we do not have.
In reality, most models will be tasked with scoring unusual, anomalous, or unexpected data at some point after they are deployed. But the more certain we are of expected conditions at the time of prediction, the more we will trust that prediction. If the system itself tells us how sure or unsure it is, we can start to think of it as a humble system, not an arrogant system, and trust it even more.
Watch the demo as Richard Tomlinson, Senior Director of Product Marketing at DataRobot, walks through the features that make up the core of Humble AI:
- Starting with a model deployed in DataRobot’s MLOps making real-time predictions, we define ‘Humble Triggers’ - situations that signal to the model that it shouldn’t feel confident. You then teach the model when it should be humble.
- Then, you define how you want the model to behave in response to this humility. What do you want it to do differently? It’s not enough to have insights - you have to take action. And DataRobot allows you to do so in an automated way.
- Finally, humility isn’t a one-off event. It’s indicative of your model being challenged in the real world. DataRobot allows you to automatically monitor your humble triggers and actions over time to ensure that your triggers and actions continue to make sense - and as time goes on you revise them.