diff --git a/awx/main/models/jobs.py b/awx/main/models/jobs.py
index d573d1ed96..19496c321c 100644
--- a/awx/main/models/jobs.py
+++ b/awx/main/models/jobs.py
@@ -900,6 +900,9 @@ class LaunchTimeConfigBase(BaseModel):
                         data[prompt_name] = self.display_extra_vars()
                     else:
                         data[prompt_name] = self.extra_vars
+                    # Depending on model, field type may save and return as string
+                    if isinstance(data[prompt_name], str):
+                        data[prompt_name] = parse_yaml_or_json(data[prompt_name])
                 if self.survey_passwords and not display:
                     data['survey_passwords'] = self.survey_passwords
             else:
diff --git a/awx/main/tests/unit/models/test_workflow_unit.py b/awx/main/tests/unit/models/test_workflow_unit.py
index f101168a8b..5f53d97a0b 100644
--- a/awx/main/tests/unit/models/test_workflow_unit.py
+++ b/awx/main/tests/unit/models/test_workflow_unit.py
@@ -234,6 +234,14 @@ class TestWorkflowJobNodeJobKWARGS:
         job_node_no_prompts.unified_job_template = project_unit
         assert job_node_no_prompts.get_job_kwargs() == self.kwargs_base
 
+    def test_extra_vars_node_prompts(self, wfjt_node_no_prompts):
+        wfjt_node_no_prompts.extra_vars = {'foo': 'bar'}
+        assert wfjt_node_no_prompts.prompts_dict() == {'extra_vars': {'foo': 'bar'}}
+
+    def test_string_extra_vars_node_prompts(self, wfjt_node_no_prompts):
+        wfjt_node_no_prompts.extra_vars = '{"foo": "bar"}'
+        assert wfjt_node_no_prompts.prompts_dict() == {'extra_vars': {'foo': 'bar'}}
+
 
 def test_get_ask_mapping_integrity():
     assert list(WorkflowJobTemplate.get_ask_mapping().keys()) == ['extra_vars', 'inventory', 'limit', 'scm_branch']