We use the classic Watson Assistant and in the past few weeks, we noticed that the intent recognition has drastically worsened in all our bots. Questions and utterances that used to have a confidence of > 0.90 now only get around 0.4 - 0.6. And that also applies to example phrases that are in the respective intent verbatim!

Has anyone else experienced something similar?

Hi Marlene,

I've noticed a worsening of intent recognition for some intents, although not necessarily all, and certainly not with the example phrases themselves. I'm also curious to hear from others on this.  Thanks,

-Patricia