The Quest for AI-Powered Fashion Advice
Navigating the unpredictable weather of March in the US can be a real challenge. It’s that awkward time of year when winter is fading, but spring hasn’t quite sprung. Your wardrobe is in limbo, with sweaters, short sleeves, jeans, and shorts all vying for attention. It’s enough to make you want to stay in your pajamas all day! But what if artificial intelligence could lend a helping hand in making those tough daily outfit decisions?
As I prepared for another day of working from home, I dutifully checked the next day’s weather forecast. For this task, I’ve always relied on Alexa, Amazon’s trusty digital assistant. While Alexa hasn’t yet received its much-anticipated AI upgrade, it has consistently served as a reliable weather reporter. So, when I inquired about the forecast for my area, Alexa presented me with a classic March dilemma: a low of 48°F and a high of 58°F, with rain throughout the day.
Naturally, I hesitated, unsure of the appropriate attire for such conditions. Was it sweater weather? T-shirt weather? Or perhaps a layered approach? It dawned on me that AI might offer a solution.
However, fashion is a deeply personal matter, and I was skeptical that generic AI advice would be sufficient. Ideally, I wanted a generative AI that could analyze my closet and, if not select an outfit outright, at least assist me in making a choice. I envisioned a system that could understand my personal style, the specific items in my wardrobe, and the nuances of the day’s weather to provide truly personalized recommendations. The challenge, I realized, was bridging the gap between general weather advice and the highly specific, subjective nature of fashion choices.
Gemini Live: A Promising Start, but Limited Vision
Since Gemini Live was readily accessible on my iPhone lock screen, and I’d previously had some success with it in game generation, I decided to begin my AI fashion experiment with Gemini, utilizing the Flash model 2.0. My reasoning was simple: Gemini Live was conveniently located and had shown some aptitude for understanding context in previous interactions.
My voice prompt was straightforward:
“If I show you my closet, can you tell me what to wear tomorrow? The high is going to be 58° the lowest going to be 48 and it’s going to be rainy.”
Gemini’s response, while prompt, was somewhat underwhelming: “Unfortunately, I can’t see what’s in your closet yet.” This immediately highlighted a key limitation: Gemini Live, at least in this version, lacked the visual processing capabilities necessary to analyze my wardrobe directly. It couldn’t ‘see’ the clothes I had available, which was a crucial part of my ideal AI fashion assistant.
Nevertheless, it did provide some sensible clothing recommendations:
“But with those temperatures, you’ll want to layer up for sure. A good base layer, like a long-sleeved t-shirt or a light sweater, is a must. On top of that, think about a jacket or coat that can keep you dry in the rain. You might also want to pack a scarf or gloves, just in case.”
Given that I hadn’t specified whether I’d be leaving the house, the suggestion of a scarf and gloves was reasonable. It demonstrated an understanding of the weather conditions and the need for protection from the elements.
While this advice was helpful, it was generic. It didn’t feel personalized or tailored to my specific situation. I was hoping for a more multimodal AI experience – one that could simultaneously process visual and auditory information about my needs, preferences, and available options to arrive at an optimal recommendation. Without that level of integration, the advantage of using AI in the first place seemed questionable. The advice, while sound, was something I could have easily figured out myself with a quick glance at the weather forecast.
Siri: A Comical Disappointment
Next, I turned to Siri, Apple’s built-in virtual assistant. I activated Siri on my iPhone 16 Pro Max and repeated my prompt: “If I show you my closet, can you tell me what to wear tomorrow? The high is going to be 58°, the lowest going to be 48°, and it’s going to be rainy.” I had hoped that Siri, being integrated into the Apple ecosystem, might offer a more seamless or insightful experience.
The outcome was hilariously inadequate: Siri simply displayed the next day’s weather forecast and offered absolutely no fashion guidance. It completely ignored the core of my request, which was to receive advice on what to wear. It’s clear that Siri is in dire need of a significant upgrade, particularly in its ability to understand and respond to complex, context-dependent queries. Siri’s failure highlighted the difference between a simple voice assistant that can retrieve information and a true AI assistant capable of providing personalized recommendations.
ChatGPT 4o: A Willing, if Imperfect, Fashion Consultant
Finally, I engaged ChatGPT 4o’s voice mode and posed essentially the same question. I was curious to see if OpenAI’s latest model would offer a more sophisticated approach to the problem. In contrast to Gemini, ChatGPT readily agreed to assist:
“Of course! Turn on your camera and show me your closet. I’ll help you pick something suitable for the weather.”
This was a significant step up from both Gemini Live and Siri. ChatGPT 4o immediately demonstrated its willingness to engage with the visual aspect of my request, indicating a multimodal capability that the other assistants lacked.
I activated my camera, began scanning the contents of my closet with my iPhone, and inquired, “What do you think?” This felt much closer to the interactive experience I had envisioned – an AI that could actually ‘see’ my clothes and offer advice based on what it saw.
ChatGPT responded:
“It looks like you have some nice long-sleeve shirts and some jackets. Since it’s going to be cool and rainy, I’d suggest wearing a long-sleeve shirt with a light jacket or sweater. Pair that with some comfortable pants, and you’ll be good to go! Do you have a favorite sweater or jacket you want to use?”
We engaged in a brief back-and-forth. At one point, I expressed my reluctance to wear a sweater (a personal protest against the stubbornly chilly weather), and ChatGPT promptly suggested a jacket as an alternative. This showed a degree of adaptability and responsiveness that was encouraging. It wasn’t just providing a pre-programmed response; it was engaging in a conversation and adjusting its recommendations based on my feedback.
Neither ChatGPT nor Gemini Live selected a specific shirt and pants combination, but I suspect my closet wasn’t arranged optimally for them to discern individual garments. A more organized closet, perhaps with items clearly visible and spaced apart, might have yielded better results. However, I appreciated that neither AI steered me toward short sleeves or a single layer, demonstrating a concern for my comfort. They both understood the basic parameters of the weather and provided advice that was generally appropriate.
The Color Blindness Challenge: Can AI Help with Matching?
To fully appreciate my fashion predicament, it’s important to understand another factor: I’m color blind. While not completely devoid of color vision, I struggle with hues, shades, and colors that are too similar, particularly reds, browns, dark blues, and dark greens. In the past, I would rely on my wife to assess my outfits before I left the house: “Do I look OK? Does this match?” A subtle shake of her head would send me back to the closet. This reliance on another person’s visual perception was a constant source of minor anxiety, and I was curious to see if AI could provide a reliable alternative.
Could my AI fashion assistants provide similar support? Could they analyze the colors of my clothing and tell me whether they matched, even if I couldn’t see the subtle differences myself?
The following morning, I gathered the jeans and long-sleeve shirt I had chosen, opened Gemini, selected the “+” icon to add a photo, and asked:
“Do you remember when we were talking yesterday about what to wear? Do these two pieces of clothing go together?”
This was a test of Gemini’s ability to maintain context across multiple interactions and to apply its visual processing capabilities to a specific fashion problem.
Gemini responded enthusiastically:
“Absolutely! It’s great to revisit this.
Yes, based on the image you’ve provided, the blue plaid shirt and the blue jeans definitely go together. Here’s why:
- Color Harmony: The blues in both pieces are complementary. The shirt has a mix of navy and lighter blues, while the jeans are a classic medium-wash blue. This creates a cohesive and visually appealing look.
- Casual Style: Both the shirt and jeans are inherently casual pieces. They work well together for a relaxed, everyday outfit.
- Versatility: The plaid shirt can be dressed up slightly with chinos or dress pants, or kept casual with jeans. Similarly, the jeans can be paired with a t-shirt, sweater, or even a blazer.
In short, this is a classic and stylish casual outfit combination!
Let me know if you’d like suggestions on how to accessorize this look or if you have any other clothing questions! 😊”
Interestingly, Gemini described my shirt as “blue plaid,” whereas I perceived it as black. This is a common occurrence for me due to my color blindness, but at least I managed to assemble an outfit with “a cohesive and visually appealing look.“ Gemini’s detailed explanation of why the clothes matched was particularly helpful. It didn’t just say “yes” or “no”; it provided a rationale based on color harmony and style, which gave me confidence in its assessment.
When I switched to ChatGPT 4o, it was equally eager to resume our fashion discussion and offered me the option of holding up my outfit or describing it. This flexibility in input methods was a nice touch.
I chose the camera button within the ChatGPT voice interface, took a picture, and promptly received a fashion assessment:
“Yes, I can see them. That plaid shirt and jeans look great together! It’s a casual and comfortable combination.”
ChatGPT didn’t comment on the specific colors of the shirt or jeans, but it seemed satisfied that I had created a well-matched ensemble. While it lacked the detailed explanation provided by Gemini, it still provided a quick and affirmative answer to my question.
A Glimpse into the Future of AI-Assisted Fashion
While this experience fell short of having a true AI fashion guru residing in my closet, one that would be intimately familiar with my wardrobe and the external weather conditions, and then curate the perfect outfit for the day, I was nonetheless impressed. ChatGPT 4o, in particular, demonstrated a more fluid and multimodal approach to fashion assistance. It was prepared to visually assess my closet, take in the overall picture, and offer at least general recommendations. The ability to seamlessly switch between voice interaction and visual input was a key differentiator.
Gemini Live, with some prompting, was not far behind and provided more detailed feedback. Therefore, ChatGPT wins by a stylish margin, but Gemini is arguably also ready for the fashion runway. The potential for AI to revolutionize personal styling is becoming increasingly apparent. The ability to integrate visual and contextual information, combined with an understanding of individual preferences, opens up exciting possibilities for creating personalized and efficient fashion recommendations.
Imagine a future where your AI assistant not only knows your entire wardrobe but also tracks your style preferences, understands your schedule (knowing whether you have important meetings or casual outings), and even takes into account your body type and skin tone. This AI could then proactively suggest outfits, taking the guesswork and stress out of getting dressed each day. It could even help you shop for new clothes, suggesting items that would complement your existing wardrobe and fit your personal style.
For individuals with color blindness or other visual impairments, the benefits could be even more profound. AI could provide a reliable and objective way to ensure that outfits are color-coordinated, eliminating the need to rely on others for visual confirmation.
While the technology is still evolving, the early results are promising, suggesting a future where AI plays a significant role in helping us all look our best, regardless of the weather or our personal color perception challenges. The development is moving faster than a runway model on a catwalk, and I believe we’re just beginning to see the full potential of AI in the world of fashion. The convergence of computer vision, natural language processing, and personalized data is creating a powerful new tool for self-expression and confidence.