Validated data fields differs from data fields Django Rest Nested serializers

So I was playing around with serializers in django and wanted to change the names of my fields in my response when I realized my changes had not been taken in count I did some digging and saw that my validated_data differs from my data.

My goal here is to give a python object to a serializer which has different fields than the name I want to return so I used the 'source=' argument to my field.

Note that changing the name of the python object's field is not an option.

Here's the python object:

class Flow(object):
    """Just a regular python object"""
    def __init__(self, name=None, flow_uid=None, data_type=None, parent=None, container=None):
        """This has more fields than the serializer is waiting"""
        self._parent = None
        self._container = None
        self.name = name
        self.data_type = data_type
        self.flow_uid = flow_uid

And the following serializers (I am using a nested representation)
serializers.py

from rest_framework.fields import CharField, IntegerField, ListField, JSONField
from rest_framework.serializers import Serializer

class OutputSerializer(Serializer):
    uid = CharField(max_length=36)
    name = CharField(max_length=100)
    description = CharField(max_length=100)


class FlowSerializer(Serializer):
    uid = CharField(source='flow_uid', max_length=36) # I want 'uid' in my response not 'flow_uid'
    name = CharField(max_length=100)
    data_type = CharField(max_length=100)
    class Meta:
        fields = '___all___'

    def to_representation(self, instance):
        instance = super(FlowSerializer, self).to_representation(instance)
        #Here instance = OrderedDict([('uid', 'uid_value'), ('name', 'name_value'), ('data_type', 'data_value')])
        return instance


class FlowOutputSerializer(OutputSerializer):
    columns = FlowSerializer(many=True)

viewsets.py

class AddTransformationViewSet(ViewSet):
    """Handle available "actions" for BrickModel operations"""

    def list(self, request, parent_lookup_analyses: str):
        """The method I call for this test"""

        flow1 = Flow(name="name1", flow_uid='flow_uid_value1', data_type='str')
        flow2 = Flow(name="name2", flow_uid='flow_uid_value2', data_type='str')
        flow1_ser = FlowSerializer(flow1)
        flow2_ser = FlowSerializer(flow2)

        dummy_col = {
            "name": "output_name",
            "description": "output_description",
            "uid": "output_uid",
            "columns":
            [
                flow2_ser.data, # Debug: {'uid': 'flow_uid_value2', 'name': 'name2', 'data_type': 'str'}
                flow1_ser.data # Debug: {'uid': 'flow_uid_value1', 'name': 'name1', 'data_type': 'str'}
            ]
        }
        #Debug dummy_col: {'name': 'output_name', 'description': 'output_description', 'uid': 'output_uid', 'columns': [{'uid': 'flow_uid_value2', 'name': 'name2', 'data_type': 'str'}, {'uid': 'flow_uid_value1', 'name': 'name1', 'data_type': 'str'}]} 


        dummy_serializer: FlowOutputSerializer = FlowOutputSerializer(data=dummy_col)
        dummy_serializer.is_valid(raise_exception=True)
        # Debug dummy_serializer.data: {'uid': 'output_uid', 'name': 'output_name', 'description': 'output_description', 'columns': [OrderedDict([('uid', 'flow_uid_value2'), ('name', 'name2'), ('data_type', 'str')]), OrderedDict([('uid', 'flow_uid_value1'), ('name', 'name1'), ('data_type', 'str')])]}
        # Debug dummy_serializer.validated_data: OrderedDict([('uid', 'output_uid'), ('name', 'output_name'), ('description', 'output_description'), ('columns', [OrderedDict([('flow_uid', 'flow_uid_value2'), ('name', 'name2'), ('data_type', 'str')]), OrderedDict([('flow_uid', 'flow_uid_value1'), ('name', 'name1'), ('data_type', 'str')])])])
        return Response(data=dummy_serializer.validated_data, status=status.HTTP_201_CREATED)

Expected_response:

{
    ...
    "columns": [
        {
            "uid": "flow_uid_value2",
            "name": "name2",
            "data_type": "str"
        },
        {
            "uid": "flow_uid_value1",
            "name": "name1",
            "data_type": "str"
        }
    ]
}

What I get (I want 'flow_uid' to be 'uid'):

{
    ...
    "columns": [
        {
            "flow_uid": "flow_uid_value2",
            "name": "name2",
            "data_type": "str"
        },
        {
            "flow_uid": "flow_uid_value1",
            "name": "name1",
            "data_type": "str"
        }
    ]
}

Is there any particular danger in using .data in this case rather than .validated_data? What is the cause of this behavior?