agentic AI in particular is so fucking funny. i run an absolutely *tiny* indie studio and i still ask people "hey could you run me through how this works?" all the time because knowing how things work is a vital part of creating a quality product
how does that work with AI? "hey could you run me through how this works?" and people just go "idk the AI did it ¯\_(ツ)_/¯"
"you can just read the code the agent wrote"
oh fuck off. the whole idea is that agents can churn out code at way higher volumes than people can generate, and the bottleneck when people wrote the code and not "agents" was already code review, because making sense of code is harder than writing it
the only thing you've done is made the code review bottleneck so, so much worse. and this will help you be more productive... how exactly?
@eniko I'm admittedly limited in my coding experience, but I'm less worried by code generation than by error fixing.
I vaguely trust the first draft to at least sorta focus on the original intent. Every iteration where code fails to compile or gives an incorrect output, though, creates a new layer of problems to focus on, and if an LLM will do ANYTHING to look good to you (as seems to be the trend), then who knows what bullshit it'll put in there just to produce something that functions?
@eniko exactly this!
I've been trying to explain this to people as well.
But it seems "we have the ai write tests too, that we also don't read" is apparently fine.
@eniko preach! This is the biggest talking point for me. When stuff falls apart, you can absolutely ask a person "Hey, why is your code like this?" And you can get some idea as to what the problem is.
With AI, you can't. It doesn't have context behind its decisions so you not only have to find out what's wrong, you have to figure out why and do so with zero guidance or input.
@eniko I’m retired now, but my career was writing code for about three and a half decades. Absolutely agree that documenting code, making sense of code, and understanding the pattern of interaction between different developer’s code is the hard part. And planning for reuse.
In my experience managers of software development almost never got that. Invariably they saw code as some kind of monolithic, homogeneous product that could be mass produced by the cheapest supplier, measured and sliced.
@eniko and we all know how much developers like doing code reviews, right?
Turn my job into one never ending code review? I’d rather quit, to be honest.
@eniko a lot of places have started use AI to do code reviews, and started to shift around who takes fault when the AI generated code goes bad (either the person who did the prompting, or the person who did the review).
So something goes wrong and one of those two gets fired, and the process that incentivizes rubber stamping PRs is never given scrutiny.
The AI only admits a fault where its training data indicates that the most likely next sentence would be admission of a fault.
Given that it's training data is the Internet where people will regularly argue well after being shown hard proof that they are wrong, that takes a lot.
@ainmosni @eniko True, but I suspect CRA/PLD/... will change this at least for some market segments.
(I'm not a huge fan of the CAIDA; we should treat "AI" in products just like normal products; as a consumer/citizen I don't much care why they don't meet requirements. That just adds to the "oh look how special AI is" crap.)
@eniko It's even narrower than that. I don't know how a Polar Code works on FPGA, but I can read a high-level wikish summary of the *intent* of the design, and I know the basic error correction coding theory to absorb what it implies for my block of the machine.
This kind of indifference can only pass when nobody knows fully how the product gets made and also the way it's made was already full of bullshit performative work-shaped nonsense even before they began asking randstr for the files.
@eniko
^ this
I suspect this is a big part of why it's so appealing to corporate bigwigs : they don't know/understand what their teams are doing anyway, so genAI doesn't change that significantly from their perspective.
@eniko
I work at a medical device company. The factory I work in, of which the company has *a bunch*, employs well over 1000 people.
Enthusiasm for generative LLMs is directly proportional to distance from production. The C_Os are urging people to adopt it. Managers are interested. Office workers are cautiously optimistic. And we on the assembly lines are all dead set against it.
@eniko recently I've been reviewing code with the most subtle and horrible kinda mistakes- and then folks who defend it "because Claude did it"
It is three times as exhausting to review this kind of crap because the mistakes are irrational, unmotivated and seem to be there solely to make reading the code harder.
@eniko my new favourite is people who do know how it works answering that you're supposed to prompt the AI now instead of share information directly 🤮
so - ok, i see we're literally not allowed to ask questions any more, is that how it fucking is
Eniko Fox
The Seven Voyages Of Steve
mirek kratochvil
James Thomson
Peter
DCoder 🇱🇹❤🇺🇦
Abyssal Rook
Tony
0xC0DEC0DE07EA
ozeng
Gabriele Svelto
Machine Lord Zero
Daniël Franke 
Ratsnake
Marta
HP van Braam
BogusMeatFactory
Michael Ormsby
Phil Haigh
Jae
Pare
Silver Owl
George B
Lars Marowsky-Brée 😷
Cogito ergo mecagoendios
rfnix
FediThing 
Infrapink (he/his/him)
Thepi
McNeely
Wharrrrrrgarbl
Mitsunee | 光音
sabik
Ian Tindale
pintozilla
Laura
grechaw
tbsp
Alex Celeste, Princess Consort of Burnout