But on the other hand what areas/specialties/domains are less likely to be replaced in the short term? (obviously anything is possible long term)
I am entirely guessing here but probably anything that is involved in extremely tight regulations and have legal oversight. Probably things related to the FAA and aircraft navigation, especially automated decisions such as TCAS Traffic Alert and Collision Avoidance Systems that automatically control real time aircraft navigation decisions. I would not doubt that AI may contribute some code at some point but everything would have to be strictly reviewed, tested and approved by multiple humans or at least that is what I would expect or hope.
On a similar note, probably land based traffic management systems that make real time decisions about traffic flows. These systems have insane levels of regulations around them and only specific people with specific degrees, certifications and licenses are permitted to even touch them much less engineer anything to do with them.
Probably something similar for water navigation especially navigating around structures like bridges, oil derriks and related platforms.
I am specifically talking about things that the HN crowd would consider to be AI LLM, big-data based platforms. There are proprietary systems that have their own concept of intelligence and automated navigation, mitigation and defense that the defense industry utilize that are unrelated to what people here call AI.