i'd bet that above some number there will be contradictions. Things that apply to different semantic contexts, but look same on syntax level (and maybe with various levels of "syntax" and "semantic"). And debugging those is going to be nightmare - same as debugging requirements spec / verification of that
let me ask a stupid/still-ignorant question - about repeatability.
If one asks this generator/assistant same request/thing, within same initial contexts, 10 times, would it generate same result ? in different sessions and all that.
because.. if not, then it's for once-off things only..
A pretty bad comparison. If I gave you the correct answer once, it's unlikely that I'll give you a wrong answer the next time. Also, aren't computers supposed to be more reliable than us? If I'm going to use a tool that behaves just like humans, why not just use my brain instead?
yeah. Long-experienced people (not just plain-old) have grown that feeling called "warm fuzzies" [1]. And although warm-fuzzies are not in the spec, mostly, it is what makes or breaks good product/service.
i had an original 7" eeepc from 2007, running archlinux-32 from ~2017, with Xfce and all that, and few months ago updated it.. took me almost a day, going through various rabbit-holes, like 1-2 static-built pacmans and python and manually picking and combining various versions. The result was okay but somehow took more space than before (it has 4G ssd, from which i did have 2gb free, now only 1.5). But it maybe that is not old enough as machine..
Technologies: python, javascript/node/react web/html/css graphql sql nosql c++ c java sh perl make linux.. dozens other languages and stacks, Prolog to Android to Verilog to assemblers and bitmasks; data+work-flows, UI/UX, testing, reverse engineering, methodology, process, management, culture, leadership, company/team-from-scratch
Domains: financial+related, any-burocracy-automating, boards/hardware/chip-making/EDA, tools+ languages (dev or process), kiosk-likes, CAD/CAM/CAE, maps, modelling-in-general - physical or virtual, ~semantical stuff, audio and many more
About: Making software for 36+ years.. Seen quite some things.. and attitudes. Always hands-on.
Looking for.. interesting work with interesting people. Maybe, principal or staff engineer, architect, CTO, co-founder, team/tech lead, head of engineering.. or very senior programmer.. (or Mentor? is this still needed? playing Trainer?). i like fiddling with things, mentoring, and building stuff - out of nothing if needed :) Effecting change, esp. cultural. Life-coaching too.. And making good friends along. But no politics, plz.
IMO The interesting part of making software, is the people, while technicals.. are solvable. And, Fun is important 4th factor in the Functionality-Price-Deadlines Triangle.
May relocate anywhere to stay with the team, once in a while, as IMO "remote" does not work well for knowledge/ culture/ relations -building activities.
especially if you include the total refusal of that caste to take responsibility - e.g. compliance - hence looking for any possibility to outsource that, be it M$ or AWS or (now new kid on da blok) "AI"
Wish there was a "capable" first management tracking tool, that rates organisations and companies by ability to create, maintain and innovate in tech. Sort of a reverse linkedIn for company culture, but only focused on that, department by department, not culture, no ammenities, just engineering culture. Must be worth something to not be stuck for a year or two in an invisible dead end filled with tech debt and pressure.
i have been programming for 40years.. but still have not dipped in this brave new world of shakespeare-taught coding-LLMs.
IMO there's one basic difference with this new "generative" stuff.. it's not deterministic. Or not yet. All previous generations "AI" were deterministic.. but died.
Generating is not a problem. i have made medium-ish projects - say 200+kloc python/js - having 50%-70% of code generated (by other code - so you maintain that meta-code, and the "language" recipes-code it interprets) - but it has been all deterministic. If shit happens - or some change is needed, anywhere on the requirements-down-to-deployment chain - someone can eventually figure out where and what. It is reasoned. And most importantly, once done, it stays done. And if i regenerate it 1000 times, it will be same.
Did this made me redundant? Not at all. Producing software is much easier this way, the recipes are much shorter, there's less space for errors etc etc. But still - Higher abstractions are even harder to grasp than boilerplate. Which has quite a cost.. you cannot throw any newbie on it and expect results.
So, fine-tuning assembly - or manual transmission - might be gonna-be-obsolete skill, as it is not required.. except in rare conditions. But it is helpful to learn stuff. To flex your mind/body about alternatives, possibilities, shortcuts, wear-and-tear, fatigue, aha-moments and what not. And then move these as concepts, onto another domains, which are not as commoditized yet.
Another thing is.. Exupery in Land-of-people talks about technology (airplanes in his case), and how without technology, mankind workarounds/ avoids places/things that are not "friendly", like twisting roads around hellscapes. While technology cuts all straight through those - flies above all that, perfect for when it works - and turns into nightmare when it breaks right in the middle of such unfriendly "area".
all these thing are.. directions. Stuff that has emerged with practicing, empirical, not a stone-cut rules. Each may apply in 99% of cases but not at all in the other 1%.. different 1% for each one. Like, you hold a knife by the handle, not by the blade. But If the handle of a knife breaks-off, you can work with the blade alone - except very-very carefully AND you better know what you're doing.
And rules depend on values anyway. So be ware (become aware) of what the values are, in the specific spot of time and space (and organisation). And learn how to make rules out of them. (universe does not obey "laws" of physics. "laws" of physics describe universe)
Software is knowledge medium. (kind-a self-) executable one. fast-changeable.
So far it needs people to (self-)change. So it is also a communication medium.. between people in space/time.. with all the trickery it entails.
And ever since it existed, progressively more levels of it are generated by machines. But it's people who cause anything.
you think this all sound too philosophical? you're right, because it is. Philosophy is the most base science there is..
yeah, and beware of the possible mild shizophreny when trying to be n concepts /persons at the same time. Might be difficult to explain to others.
but do not be afraid.
and. last but not least :).. have fun
i'd bet that above some number there will be contradictions. Things that apply to different semantic contexts, but look same on syntax level (and maybe with various levels of "syntax" and "semantic"). And debugging those is going to be nightmare - same as debugging requirements spec / verification of that
reply