You're right to listen to those doubts.  The use of LLMs is a
cognitive hazard, magnified when used in contexts you're not already
an expert in: https://www.baldurbjarnason.com/2025/trusting-your-own-judgement-on-ai/

The code they produce, even when it functions as intended, is
impossible to maintain: code without an author is tech debt as soon as
it is produced, because there is no individual with the understanding
that comes from writing it.

Even if the above was not the case, widespread use of LLMs is an
environmental and social disaster, consuming resources at a
catastrophic rate and supporting noxious ideologies:
https://aworkinglibrary.com/writing/toolmen

Instead of engaging with LLMs, consider collaborating with other human
colleagues on sustainable, ethical work.

On Mon, Jun 8, 2026 at 6:10 PM Mark Andrews
<[log in to unmask]> wrote:
>
> I've not been a list member in some years and recently joined again. I hope this question - "Whither the ILS or LSP" - is not off-putting. I wonder if this problem has been solved to death? Here is why I ask.
>
> I received my MLS from Mizzou in May '86 and started my first library job in June '86. Fast forward and I've worked 40 years. That includes 9 years in the library automation industry and 9 more years as a systems librarian. The only thing I've never done is write a serious piece of code. Last November, out of curiosity and boredom, I started "vibe" coding. I tried out a variety LLMs before I found one I like using. It cost me $300 or so over 6 months. Not an outrageous amount of money. A first lesson is it is a rush to see the machine write code, and I can easily spend myself broke without some discipline.
>
> My tools of choice are Golang, SQL, the Rego configuration language, and the Qwen3.6 Plus model on opencode.ai.
>
> I have the bare skeleton of something remotely like a product. It's made of 37.5K lines of Golang (excluding a vast amount of generated gRPC protobuf stubs) and 1,350 of lines SQL for MariaDB. It compiles and I can load bib and authority records to my heart's content. All greenfield code based on years spent supporting 6 ILSs, large and small. It was at this point I remembered the sage advice of "Cookiness Evereat" on Sesame Street, who said, "Wait! Me gotta stop, and think!"
>
> "Do I really want to build this thing? Really?" Nine years on The Dark Side made me and broke me. When I found myself perilously close to doing things to my customers and not for them, I quit. I've felt guilty about that for decades. Does the profession truly need another tool and am I the person to build it? I do not know. I do know I can't make this decision alone. I am posting here looking for honest feedback. Truly, I may be of more use quietly advancing someone else's project under their leadership.
>
> FYI, I'm an app admin at Creighton University in Omaha, NE. I don't speak for Creighton in any way, shape or form. I'll finish 20 years there in January 2027. Thank you in advance for your time and help. I wish you all peace and Days Without Drama (tm).
>
> Mark Andrews