Explainable AI refers to the collection of methods and techniques which enable a human to understand why a model is giving specific results. This can be done in Google Cloud with the following steps:

1. Train a model and deploy it on GCP.
2. Upload a JSON file that contains baseline feature values to a Cloud Storage Bucket.
3. Create a version of the model using this JSON file and specifying the explanation-method.
4. Call gcloud beta ai-platform explain to get the explanations.

The detailed steps are found in the above notebook or in a Colab notebook.