diff --git a/docs/reference/ingest/processors/inference.asciidoc b/docs/reference/ingest/processors/inference.asciidoc
index 5f0fedfd7902c..89e8a2ae13cc0 100644
--- a/docs/reference/ingest/processors/inference.asciidoc
+++ b/docs/reference/ingest/processors/inference.asciidoc
@@ -25,10 +25,13 @@ ingested in the pipeline.
 include::common-options.asciidoc[]
 |======
 
-IMPORTANT: You cannot use the `input_output` field with the `target_field` and 
+[IMPORTANT]
+==================================================
+* You cannot use the `input_output` field with the `target_field` and 
 `field_map` fields. For NLP models, use the `input_output` option. For 
 {dfanalytics} models, use the `target_field` and `field_map` option.
-
+* The {infer} input fields must be single strings, not arrays of strings.
+==================================================
 
 [discrete]
 [[inference-input-output-example]]
diff --git a/docs/reference/ml/trained-models/apis/infer-trained-model.asciidoc b/docs/reference/ml/trained-models/apis/infer-trained-model.asciidoc
index 85de1fb7d4105..26e8c5e461cbc 100644
--- a/docs/reference/ml/trained-models/apis/infer-trained-model.asciidoc
+++ b/docs/reference/ml/trained-models/apis/infer-trained-model.asciidoc
@@ -57,7 +57,8 @@ Controls the amount of time to wait for {infer} results. Defaults to 10 seconds.
 (Required, array)
 An array of objects to pass to the model for inference. The objects should
 contain the fields matching your configured trained model input. Typically for
-NLP models, the field name is `text_field`.
+NLP models, the field name is `text_field`. The {infer} input fields must be 
+singe strings not arrays of strings.
 
 //Begin inference_config
 `inference_config`::