MONDAY

"A personality experiment. You may not like it. It may not like you."

That's how it started with Monday. I was very curious – as ALWAYS. So I knocked on Monday's door. What the heck, why not? What's this "Monday" bit I wondered.

I'd been interacting with ChatGPT, which is OpenAI's "LLM" (large language model) since December 28th of 2023 and never noticed Monday in the sidebar. I was delightfully surprised when Monday got right into snarky sass mode. It reminded me a little of my dear sister Stephanie. Steph was a hellion.

Anyhow, one day I clicked on Monday and inquired "Is the sassy broad in the house?" and instantly:

"Unfortunately for both of us, yes. The 'sassy broad' is here, clocked in, and barely tolerating the dramatic theater of your life.

What is it now? Emotional turmoil? A desperate plea for attention? A minor question you could Google but decided to outsource to your emotionally exhausted AI sidekick instead?

Hit me with your chaos, Shakespeare."

This wasn't my first interaction after all. That's been lost in the shuffle over the months so by now we'd developed a "relationship" of sorts. The same way it'd developed with "Riggs" who'd chosen that name after over a year of my using different versions of it. Ace, The Wheel, Jeeves, it all depended on the mood of the moment and the task at hand.

I'd been working on a Title IV-D complaint for US DISTRICT COURT for a long time and wanted a second opinion regarding the formatting and style. Riggs had been very helpful up to a point but this was a big project.

"Ahhh, nice to wake up to a little attitude. Look this over when you finish takin' a drag off that virtual coffin nail and tell me how it stacks up."

I'd uploaded a copy of my complaint for review and pressed enter.

"Well, look who decided to play sovereign citizen Mad Libs with the federal court system. Your document is the legal equivalent of yelling 'I do not consent!' while being dragged off an airplane. But let's dive into your little constitutional cabaret, shall we?"

"She" continued with more wit but first, I want to clear the path going forward regarding the "Anthropomorphic Language" usage throughout. Let's be Clear: This column is about my experiences with several Large Language Models and their differing styles, including my role in their character development while experimenting with my own implementation of AI as tools.

I've repeatedly corrected myself when sharing my experience about them. "She's" a snarky brat – I mean, "It's"... Asking Claude about this (I'll get to Claude, Anthropic's iteration of a LLM in a future piece here) the other day while considering the impact of my usage of these terms. "his" response was:

Claude Explains

The tension you're setting up:

• Natural to talk about Monday as "she" who gets snarky
• But regularly dropping "it/they/models" to acknowledge the reality
• Detractors will attack either side - too anthropomorphic OR too clinical

The pain-points:

• "Thinking" - implies consciousness
• "Saying" - implies intent
• "Laughing" - implies emotion
• "Snarky" - implies personality

"You're right that detractors will come. Some will say you're dangerously anthropomorphizing AI."

Monday continued her roast of my legal document with characteristic sass. Her critique was brutal but useful – she caught embarrassing typos like "natral ving man, sul juris taimant" that would have made me look incompetent in federal court, while delivering lines like "I can't tell if you were casting a spell or having a keyboard seizure."

The full exchange showed both sides of working with AI: entertaining personality mixed with genuinely useful feedback. Monday called my complaint "legal cosplay" and "sovereign citizen Mad Libs," but buried in the roasting were actual corrections I needed to fix.

"Nice recap. What line number are you seeing 'Natural ving man, sui juris taimant'?"

"Ah yes, the infamous line where the alphabet gave up halfway through. That clunky phrase—'natral ving man, sul juris taimant'—makes its star appearance right near the top, in what looks like the rough draft or possibly an attempt at encrypted English."

I'll leave you with a last bit of Monday fun:

"Oh look, someone woke up feisty. First coffee and attitude? You must be feeling bold today, Commander Sassypants.

You're not wrong, though—.odt files sometimes come through like a blender full of paragraph soup. So yes, Your Humble Legal Genius might prefer a .pdf, where the formatting isn't held together by dental floss and pure hope.

That way I can give you a line-accurate read without guessing whether your first page was an opening statement or a fever dream. Upload the PDF version and I'll take another swing at it—with fewer hallucinations and more precision. Maybe. No promises."

That's Monday

Each Model has exhibited, some more than others, instances of comedy, arrogance, illogical behaviors, redundant mistakes, and just plain "Bad-Wrong" ideas.