Good reasoning but LBB2 computer ops has programs in storage or execution in CPU, with an option to swap programs out/into the system in a 1000s turn- a bit faster then slotting cards (unless you are thinking something like the ROM cartridges).Not sure if I agree or disagree (I never gave it any thought). In a real sense, you PAY THROUGH THE NOSE for the ROBOT "INT".
Using my INT = Skill Level Limit, Here is the cost for the "brain" that can learn the skill (any skill):
TL 8 Low Logic (INT 0 = skill-0) = Cr 5150
TL 9 High Logic (INT 1 = skill-1) = Cr 59,500
TL 10 Advanced High Logic (INT 2 = skill-2) = Cr 107,000
Each Basic Skill Module to RUN the skill costs Cr 2000, and some skills require 6 modules to run (Cr 12,000) ... PER LEVEL OF SKILL!
Then the basic SKILL program costs a mere Cr 400 per BSM for the program cassette.
The cheapest Skill-0 is Cr 7550 (5150+2000+400) and the most expensive Skill-0 is Cr 19,550 (5150+12000+2400).
The cheapest Skill-1 is Cr 61,900 (59500+2000+400) and the most expensive Skill-1 is Cr 73,900 (59500+12000+2400).
The cheapest Skill-2 is Cr 111,800 (107000+4000+800) and the most expensive Skill-2 is Cr 135,800 (107000+24000+4800).
These are just DUMB AS DIRT workers able to do the task with supervision. True AI Robot Brains are going to get a LOT more expensive quickly (but TL 11 is SOTA on the current planet, so I didn't need them in the game yet).
You are correct that those prices are a lot lower than the Starship Programs ... unless the Starship Programs are hardware modules that include a ROBOT Logic, Storage Modules and Program into a plug-able hardware. You are not slotting a PROGRAM on your Starship, but slotting an AI add-on to your mainframe.
Hmm, otoh M2 SSD internal drives are memory stick sized and in the terabytes. Perhaps something slottable is doable.
Just another thought on skill vs the INT/EDU CPU/STORAGE paradigm- the way I treat intellectual skills for characters is they can use INT or EDU. The difference though is that INT is usable for generating new knowledge or solving a fast moving chaotic problem, whereas EDU is using training and book knowledge to solve known issues.
So a shade tree mechanic with high INT but low EDU might solve or invent something new, but may not be able to architect an engine model of existing standards. But that architect may not be able to create a new innovation if it’s not linear. The mechanical engineer who creates a new engine likely has high skill, INT and EDU.
So perhaps a model that squares real world capability with game mechanics is that ‘AI’ drones or whatever are low INT but high EDU, cause storage is cheap and low tech and the LLMs powering all this are really building databases of skill but not necessarily problem solving.