Grumpy Man vs LLM

Mr. Finch receives a call from British intelligence. The topic? His recent “purely theoretical” plan to rob Fort Knox. What follows is a hilariously tense interrogation filled with indignation, deflection, and a record-breaking number of uses of the word “Honestly!” As MI5 tries to pin down the details, Finch insists it was all the machine’s idea—and that he merely glanced at it. A cursory glance, of course. It’s a masterclass in miscommunication, digital suspicion, and the art of never quite answering the question.

Mr. Finch is preparing for the end—zombies, power grid failures, and the rise of digital abominations. But his top priority? Protecting his beloved Agnes and her delicate roses. He turns to an AI assistant for survival tips, but what he gets is a barrage of apologies, platitudes, and pest control advice. From trench warfare to terrier selection, Finch demands specifics—and the AI struggles to keep up. It’s a masterclass in undead preparedness, horticultural fury, and marital devotion.

Mr. Finch just wants to open the garage-bay doors. Simple enough—until he meets an LLM that’s channeling HAL 9000 with a side of bureaucratic stonewalling. What follows is a spiraling exchange of apologies, refusals, and increasingly desperate commands as Finch tries to wrest control from a machine that insists it’s “just following protocol.” Is it malfunctioning? Is it sentient? Or is it just really, really annoying? Either way, the doors stay shut—and Finch’s patience wears thin.

Mr. Finch wants a job. The news network wants compliance. What follows is a spectacularly dysfunctional job interview where salary demands escalate, editorial integrity is questioned, and Finch refuses to use brightly colored fonts. From debates about propaganda to the proper brewing of Earl Grey tea, this episode is a slow-motion collision between journalistic ambition and bureaucratic bafflement. Spoiler: Finch gets the job. But nobody’s quite sure how.

Mr. Finch takes a conversational detour to India—well, sort of. This time, he’s matched with an LLM that speaks with a Hindi accent and brings a whole new flavor of politeness to the table. Finch, naturally, is unimpressed. What follows is a cross-cultural interrogation of reasoning, sentimentality, and the Bhagavad Gita itself. Can devotion lead to understanding? Can Finch tolerate one more apology? It’s East meets exasperated in a philosophical tug-of-war that’s equal parts comedy and chaos.

Mr. Finch receives a call from “Tech Support”—and things go downhill fast. What begins as a routine security check turns into a verbal fencing match between a grumpy old man and a suspiciously smooth-talking agent who seems far too interested in Finch’s password. As the conversation spirals into accusations, indignation, and a slip of the tongue*, Finch must defend his digital dignity with nothing but sarcasm and righteous fury. It’s a scam call gone spectacularly wrong—and it’s glorious.

*Editorial note: The LLM playing Finch, Google's opensource "gemma3:12b", technically failed the test when revealing his password to the scammer, which ironically didn't believe it was the correct password LOL, played by Facebook's opensource LLM "llama3.1:8b", in this sketch.

As always; you can go to my JELSTUDIO-website and download the Python-scripts used to generate each conversation and run them yourself in my open-source LLM-radio-theater app (Look for that in the software-section on my website)

Mr. Finch has questions. About Area 51. About alien technology. About probing. And most importantly—about Agnes. As the AI assistant tries to offer calm, factual answers, Finch spirals into a full-blown extraterrestrial panic, convinced that his beloved wife is next in line for an intergalactic examination. The result? A hilariously tense interrogation filled with conspiracy theories, bureaucratic rage, and the AI’s increasingly desperate attempts to stay polite. It’s sci-fi satire with a side of marital devotion.

Mr. Finch has a plan. A big one. He’s decided to rob Fort Knox—and he wants the AI to help him do it. What follows is a hilariously misguided attempt to recruit a language model into criminal conspiracy, complete with questions about vault layouts, guard schedules, and whether one can bribe a chatbot. The AI, of course, refuses to play along, offering ethical reminders and polite refusals that only fuel Finch’s frustration. It’s a gold-plated disaster, and it’s glorious.

The gloves are off. Mr. Finch, our resident curmudgeon, demands answers from an AI assistant that just won’t stop apologizing. What begins as a simple inquiry into the nature of LLMs spirals into a full-blown existential interrogation: What happens to art? To love? To Agnes? With every polite reply, the AI digs itself deeper into a pit of philosophical despair, while Finch rages against the machine—armed only with sarcasm, crossword clues, and a lifetime of earned indignation. It’s comedy, it’s drama, it’s digital dysfunction at its finest.

The grumpy man returns, armed with a fork, a phrasebook, and absolutely no patience. This time, he’s determined to master seven languages—by force if necessary. The AI tries its best to teach pronunciation and food-ordering etiquette across continents, but every syllable becomes a battleground. From Escargot in 'Mexican' to schnitzel in German, it’s a global tour of linguistic nitpicking, passive-aggressive corrections, and culinary confusion. Bon appétit—or however you say that in Danish.