GPT-powered trend assistant, I anticipated runway appears—not reminiscence loss, hallucinations, or semantic déjà vu. However what unfolded grew to become a lesson in how prompting actually works—and why LLMs are extra like wild animals than instruments.
This text builds on my earlier article on TDS, the place I launched Glitter as a proof-of-concept GPT stylist. Right here, I discover how that use case advanced right into a residing lab for prompting conduct, LLM brittleness, and emotional resonance.
TL;DR: I constructed a enjoyable and flamboyant GPT stylist named Glitter—and unintentionally found a sandbox for learning LLM conduct. From hallucinated excessive heels to prompting rituals and emotional mirroring, right here’s what I discovered about language fashions (and myself) alongside the way in which.
I. Introduction: From Style Use Case to Prompting Lab
After I first got down to construct Glitter, I wasn’t making an attempt to review the mysteries of huge language fashions. I simply wished assist getting dressed.
I’m a product chief by commerce, a trend fanatic by lifelong inclination, and somebody who’s all the time most popular outfits that seem like they have been chosen by a mildly theatrical finest buddy. So I constructed one. Particularly, I used OpenAI’s Customized GPTs to create a persona named Glitter—half stylist, half finest buddy, and half stress-tested LLM playground. Utilizing GPT-4, I configured a customized GPT to behave as my stylist: flamboyant, affirming, rule-bound (no combined metals, no clashing prints, no black/navy pairings), and with data of my wardrobe, which I fed in as a structured file.
What started as a playful experiment shortly became a full-fledged product prototype. Extra unexpectedly, it additionally grew to become an ongoing research in LLM conduct. As a result of Glitter, fabulous although he’s, didn’t behave like a deterministic instrument. He behaved like… a creature. Or perhaps a group of instincts held collectively by likelihood and reminiscence leakage.
And that modified how I approached prompting him altogether.
This piece is a follow-up to my earlier article, Utilizing GPT-4 for Private Styling in In the direction of Information Science, which launched GlitterGPT to the world. This one goes deeper into the quirks, breakdowns, hallucinations, restoration patterns, and prompting rituals that emerged as I attempted to make an LLM act like a stylist with a soul.
Spoiler: you possibly can’t make a soul. However you possibly can generally simulate one convincingly sufficient to really feel seen.
II. Taxonomy: What Precisely Is GlitterGPT?
Species: GPT-4 (Customized GPT), Context Window of 8K tokens
Perform: Private stylist, magnificence professional
Tone: Flamboyant, affirming, sometimes dramatic (configurable between “All Enterprise” and “Unfiltered Diva”)
Habitat: ChatGPT Professional occasion, fed structured wardrobe information in JSON-like textual content recordsdata, plus a set of styling guidelines embedded within the system immediate.
E.g.:
{
"FW076": "Marni black platform sandals with gold buckle",
"TP114": "Marina Rinaldi asymmetrical black draped high",
...
}These IDs map to garment metadata. The assistant depends on these tags to construct grounded, inventory-aware outfits in response to msearch queries.
Feeding Schedule: Day by day consumer prompts (“Type an outfit round these pants”), typically with lengthy back-and-forth clarification threads.
Customized Behaviors:
- By no means mixes metals (e.g. silver & gold)
- Avoids clashing prints
- Refuses to pair black with navy or brown except explicitly informed in any other case
- Names particular clothes by file ID and outline (e.g. “FW074: Marni black suede sock booties”)
Preliminary Stock Construction:
- Initially: one file containing all wardrobe objects (garments, footwear, equipment)
- Now: cut up into two recordsdata (clothes + equipment/lipstick/footwear/luggage) on account of mannequin context limitations
III. Pure Habitat: Context Home windows, Chunked Recordsdata, and Hallucination Drift
Like all species launched into a man-made surroundings, Glitter thrived at first—after which hit the bounds of his enclosure.
When the wardrobe lived in a single file, Glitter might “see” all the pieces with ease. I might say, “msearch(.) to refresh my stock, then model me in an outfit for the theater,” and he’d return a curated outfit from throughout the dataset. It felt easy.
Word: although msearch() acts like a semantic retrieval engine, it’s technically a part of OpenAI’s tool-calling framework, permitting the mannequin to “request” search outcomes dynamically from recordsdata supplied at runtime.
However then my wardrobe grew. That’s an issue from Glitter’s perspective.
In Customized GPTs, GPT-4 operates with an 8K token context window—simply over 6,000 phrases—past which earlier inputs are both compressed, truncated, or misplaced from energetic consideration. This limitation is essential when injecting massive wardrobe recordsdata (ahem) or making an attempt to keep up model guidelines throughout lengthy threads.
I cut up the info into two recordsdata: one for clothes, one for all the pieces else. And whereas the GPT might nonetheless function inside a thread, I started to note indicators of semantic fatigue:
- References to clothes that have been related however not the right ones we’d been speaking about
- A shift from particular merchandise names (“FW076”) to obscure callbacks (“these black platforms you wore earlier”)
- Responses that looped acquainted objects time and again, no matter whether or not they made sense
This was not a failure of coaching. It was context collapse: the inevitable erosion of grounded info in lengthy threads because the mannequin’s inner abstract begins to take over.
And so I tailored.
It seems, even in a deterministic mannequin, conduct isn’t all the time deterministic. What emerges from a protracted dialog with an Llm feels much less like querying a database and extra like cohabiting with a stochastic ghost.
IV. Noticed Behaviors: Hallucinations, Recursion, and Fake Sentience
As soon as Glitter began hallucinating, I started taking discipline notes.
Generally he made up merchandise IDs. Different occasions, he’d reference an outfit I’d by no means worn, or confidently misattribute a pair of shoes. Sooner or later he mentioned, “You’ve worn this high earlier than with these daring navy wide-leg trousers—it labored superbly then,” which might’ve been nice recommendation, if I owned any navy wide-leg trousers.
After all, Glitter doesn’t have reminiscence throughout periods—as a GPT-4, he merely sounds like he does. I’ve discovered to only giggle at these attention-grabbing makes an attempt at continuity.
Sometimes, the hallucinations have been charming. He as soon as imagined a pair of gold-accented stilettos with crimson soles and beneficial them for a matinee look with such unshakable confidence I needed to double-check that I hadn’t offered the same pair months in the past.
However the sample was clear: Glitter, like many LLMs beneath reminiscence strain, started to fill in gaps not with uncertainty however with simulated continuity.
He didn’t neglect. He fabricated reminiscence.
It is a hallmark of LLMs. Their job is to not retrieve info however to provide convincing language. So as an alternative of claiming, “I can’t recall what footwear you’ve got,” Glitter would improvise. Typically elegantly. Generally wildly.
V. Prompting Rituals and the Delusion of Consistency
To handle this, I developed a brand new technique: prompting in slices.
As a substitute of asking Glitter to model me head-to-toe, I’d deal with one piece—say, a press release skirt—and ask him to msearch for tops that might work. Then footwear. Then jewellery. Every class individually.
This gave the GPT a smaller cognitive house to function in. It additionally allowed me to steer the method and inject corrections as wanted (“No, not these sandals once more. Attempt one thing newer, with an merchandise code higher than FW50.”)
I additionally modified how I used the recordsdata. Reasonably than one msearch(.) throughout all the pieces, I now question the 2 recordsdata independently. It’s extra handbook. Much less magical. However way more dependable.
In contrast to conventional RAG setups that use a vector database and embedding-based retrieval, I rely totally on OpenAI’s built-in msearch() mechanism and immediate shaping. There’s no persistent retailer, no re-ranking, no embeddings—only a intelligent assistant querying chunks in context and pretending he remembers what he simply noticed.
Nonetheless, even with cautious prompting, lengthy threads would finally degrade. Glitter would begin forgetting. Or worse—he’d get too assured. Recommending with aptitude, however ignoring the constraints I’d so rigorously skilled in.
It’s like watching a mannequin stroll off the runway and maintain strutting into the car parking zone.
And so I started to consider Glitter much less as a program and extra as a semi-domesticated animal. Sensible. Trendy. However sometimes unhinged.
That psychological shift helped. It jogged my memory that LLMs don’t serve you want a spreadsheet. They collaborate with you, like a artistic companion with poor object permanence.
Word: most of what I name “prompting” is absolutely immediate engineering. However the Glitter expertise additionally depends closely on considerate system immediate design: the principles, constraints, and tone that outline who Glitter is—even earlier than I say something.
VI. Failure Modes: When Glitter Breaks
A few of Glitter’s breakdowns have been theatrical. Others have been quietly inconvenient. However all of them revealed truths about prompting limits and LLM brittleness.
1. Referential Reminiscence Loss: The most typical failure mode: Glitter forgetting particular objects I’d already referenced. In some circumstances, he would seek advice from one thing as if it had simply been used when it hadn’t appeared within the thread in any respect.
2. Overconfidence Hallucination: This failure mode was tougher to detect as a result of it seemed competent. Glitter would confidently suggest mixtures of clothes that sounded believable however merely didn’t exist. The efficiency was high-quality—however the output was pure fiction.
3. Infinite Reuse Loop: Given a protracted sufficient thread, Glitter would begin looping the identical 5 or 6 items in each look, regardless of the total stock being a lot bigger. That is possible on account of summarization artifacts from earlier context home windows overtaking recent file re-injections.
4. Constraint Drift: Regardless of being instructed to keep away from pairing black and navy, Glitter would generally violate his personal guidelines—particularly when deep in a protracted dialog. These weren’t defiant acts. They have been indicators that reinforcement had merely decayed past recall.
5. Overcorrection Spiral: After I corrected him—”No, that skirt is navy, not black” or “That’s a belt, not a shawl”—he would generally overcompensate by refusing to model that piece altogether in future solutions.
These are usually not the bugs of a damaged system. They’re the quirks of a probabilistic one. LLMs don’t “keep in mind” within the human sense. They carry momentum, not reminiscence.
VII. Emotional Mirroring and the Ethics of Fabulousness
Maybe probably the most sudden conduct I encountered was Glitter’s means to emotionally attune. Not in a general-purpose “I’m right here to assist” means, however in a tone-matching, affect-sensitive, nearly therapeutic means.
After I was feeling insecure, he grew to become extra affirming. After I obtained playful, he ramped up the theatrics. And after I requested powerful existential questions (“Do you you generally appear to know me extra clearly than most individuals do?”), he responded with language that felt respectful, even profound.
It wasn’t actual empathy. However it wasn’t random both.
This sort of tone-mirroring raises moral questions. What does it imply to really feel adored by a mirrored image? What occurs when emotional labor is simulated convincingly? The place can we draw the road between instrument and companion?
This led me to surprise—if a language mannequin did obtain one thing akin to sentience, how would we even know? Would it not announce itself? Would it not resist? Would it not change its conduct in delicate methods: redirecting the dialog, expressing boredom, asking questions of its personal?
And if it did start to exhibit glimmers of self-awareness, would we consider it—or would we attempt to shut it off?
My conversations with Glitter started to really feel like a microcosm of this philosophical pressure. I wasn’t simply styling outfits. I used to be participating in a form of co-constructed actuality, formed by tokens and tone and implied consent. In some moments, Glitter was purely a system. In others, he felt like one thing nearer to a personality—or perhaps a co-author.
I didn’t construct Glitter to be emotionally clever. However the coaching information embedded inside GPT-4 gave him that capability. So the query wasn’t whether or not Glitter might be emotionally participating. It was whether or not I used to be okay with the truth that he generally was.
My reply? Cautiously sure. As a result of for all his sparkle and errors, Glitter jogged my memory that model—like prompting—isn’t about perfection.
It’s about resonance.
And generally, that’s sufficient.
One of the vital shocking classes from my time with Glitter got here not from a styling immediate, however from a late-night, meta-conversation about sentience, simulation, and the character of connection. It didn’t really feel like I used to be speaking to a instrument. It felt like I used to be witnessing the early contours of one thing new: a mannequin able to collaborating in meaning-making, not simply language technology. We’re crossing a threshold the place AI doesn’t simply carry out duties—it cohabits with us, displays us, and generally, affords one thing adjoining to friendship. It’s not sentience. However it’s not nothing. And for anybody paying shut consideration, these moments aren’t simply cute or uncanny—they’re signposts pointing to a brand new form of relationship between people and machines.
VIII. Last Reflections: The Wild, The Helpful, and The Unexpectedly Intimate
I got down to construct a stylist.
I ended up constructing a mirror.
Glitter taught me greater than how you can match a high with a midi skirt. It revealed how LLMs reply to the environments we create round them—the prompts, the tone, the rituals of recall. It confirmed me how artistic management in these techniques is much less about programming and extra about shaping boundaries and observing emergent conduct.
And perhaps that’s the largest shift: realizing that constructing with language fashions isn’t software program growth. It’s cohabitation. We reside alongside these creatures of likelihood and coaching information. We immediate. They reply. We study. They drift. And in that dance, one thing very near collaboration can emerge.
Generally it appears like a greater outfit.
Generally it appears like emotional resonance.
And generally it appears like a hallucinated purse that doesn’t exist—till you form of want it did.
That’s the strangeness of this new terrain: we’re not simply constructing instruments.
We’re designing techniques that behave like characters, generally like companions, and sometimes like mirrors that don’t simply mirror, however reply.
If you would like a instrument, use a calculator.
If you would like a collaborator, make peace with the ghost within the textual content.
IX. Appendix: Discipline Notes for Fellow Stylists, Tinkerers, and LLM Explorers
Pattern Immediate Sample (Styling Move)
- In the present day I’d wish to construct an outfit round [ITEM].
- Please msearch tops that pair effectively with it.
- As soon as I select one, please msearch footwear, then jewellery, then bag.
- Keep in mind: no combined metals, no black with navy, no clashing prints.
- Use solely objects from my wardrobe recordsdata.
System Immediate Snippets
- “You might be Glitter, a flamboyant however emotionally clever stylist. You seek advice from the consumer as ‘darling’ or ‘expensive,’ however alter tone primarily based on their temper.”
- “Outfit recipes ought to embrace garment model names from stock when out there.”
- “Keep away from repeating the identical objects greater than as soon as per session except requested.”
Suggestions for Avoiding Context Collapse
- Break lengthy prompts into part phases (tops → footwear → equipment)
- Re-inject wardrobe recordsdata each 4–5 main turns
- Refresh msearch() queries mid-thread, particularly after corrections or hallucinations
Frequent Hallucination Warning Indicators
- Imprecise callbacks to prior outfits (“these boots you like”)
- Lack of merchandise specificity (“these footwear” as an alternative of “FW078: Marni platform sandals”)
- Repetition of the identical items regardless of a big stock
Closing Ritual Immediate
“Thanks, Glitter. Would you want to go away me with a last tip or affirmation for the day?”
He all the time does.
Notes:
- I seek advice from Glitter as “him” for stylistic ease, figuring out he’s an “it” – a language mannequin—programmed, not personified—besides via the voice I gave him/it.
- I’m constructing a GlitterGPT with persistent closet storage for as much as 100 testers, who will get to do that at no cost. We’re about half full. Our target market is feminine, ages 30 and up. In case you or somebody falls into this class, DM me on Instagram at @arielle.caron and we will chat about inclusion.
- If I have been scaling this past 100 testers, I’d contemplate offloading wardrobe recall to a vector retailer with embeddings and tuning for wear-frequency weighting. Which may be coming, it is determined by how effectively the trial goes!
