Ok, if this date format comes from a CSV, you have to convert it.
parseDateTime(
outputs('Split_Each_Item')[1]
)
This is how I would build it (simplyfied 😉).
It avoids the loop, which slows down the flow and saves some actions, making it easier to keep track of.

Here the CSV data is in a Compose action
Name,Created,Start
Jim,2023-08-13 00:01:12 ,2023-09-02 01:00:00
Jim,2023-08-14 00:01:12 ,2023-09-03 01:00:00
Jim,2023-08-15 00:01:12 ,2023-09-04 01:00:00
Select
From:
skip(
split(outputs('Compose'), decodeUriComponent('%0A')),
1
)
Map
Name:
split(item(), ',')[0]
Created:
parseDateTime(split(item(), ',')[1])
Start:
parseDateTime(split(item(), ',')[2])
Diff:
dateDifference(
parseDateTime(split(item(), ',')[1]),
parseDateTime(split(item(), ',')[2])
)
Output
[
{
"Name": "Jim",
"Created": "2023-08-13T00:01:12.0000000",
"Start": "2023-09-02T01:00:00.0000000",
"Diff": "20.00:58:48"
},
{
"Name": "Jim",
"Created": "2023-08-14T00:01:12.0000000",
"Start": "2023-09-03T01:00:00.0000000",
"Diff": "20.00:58:48"
},
{
"Name": "Jim",
"Created": "2023-08-15T00:01:12.0000000",
"Start": "2023-09-04T01:00:00.0000000",
"Diff": "20.00:58:48"
}
]
'