Paraphrasing a real conversation I had.
GenAI Executive Acolyte (GAIEA): I've been following this thread about the report you are trying to create with disparate data coming from various sources. I was thinking we could throw genAI at it.
Me: Oh. Were you wanting to use the genAI to develop something in Python to pull in the data and mangle it into the correct formats and then perhaps creating nice Power Queries in Power BI I can use for reporting? Maybe create a design for some database tables based on my requirements for long term storage for trending?
GAIEA: I didn't even think of that! I thought we'd build a template of instructions and have agentic "AI" do the work autonomously every week.
Me: Wouldn't that use a lot of tokens every week? Wouldn't it be more cost effective to use genAI to design the system and then use the system directly?
GAIEA: I just really think we should burn the tokens for reasons that I cannot adequately explain other than "genAI is so cool right now".
Okay, that last bit is embellished, but only slightly. He honestly did think his way was better and didn't understand my point of view.
I cannot even begin to comprehend how they can think like this. I offered a valid use case where we only have to have our guaranteed "human in the loop" (as required) for the development cycle and then it runs safely and deterministically from then on and he countered with making it more error prone, harder to understand, and requiring far more human effort to babysit.
That's what I'm working with. There is no logic or reason, just blinken lights and shinies and dopamine hits.
I was very polite to him since he is an executive and I like my job. I do my best to keep my honest opinions close to my vest and try to be more like, "I don't really hate genAI, just how it's used in certain (most) instances and that people put an unwarranted amount of trust trust in it. Hey did you see this article about Richard Dawkins?"
But I do actually hate genAI.
#NoAI #NoGenAi