We Asked AI to Cook Vietnamese, Lebanese, and Korean From the Same 6 Ingredients

The same chicken, garlic, and soy sauce. Three completely different meals. Here's what happened.

---

There's something I've always found quietly fascinating about the fact that Vietnamese pho, Lebanese shawarma, and Korean bibimbap don't actually require that different a pantry. The fundamentals — protein, aromatics, acid, carbs, something savory — are essentially the same across cuisines. What changes is everything else: the spice combinations, the cooking technique, the relationship between those ingredients, the whole cultural framework around what a meal is supposed to feel like.

I wanted to test how far that idea actually stretched. So I ran an experiment: six ingredients, three cuisines, and Frittu's Cuisine Explorer to generate each recipe. The ingredients were deliberately ordinary — stuff that's probably already sitting in your fridge and cupboard right now. No special trips to the market, no exotic imports. Just: chicken thighs, garlic, soy sauce, lemon, jasmine rice, and a spice drawer raid (cumin, paprika, mixed dried herbs, whatever was in there).

---

The 6 Ingredients

Here's what I started with, specifically so this felt like a real test and not a setup:

• Chicken thighs (bone-in, skin on — the only honest way to buy them)
• Garlic (half a bulb, not a fresh head — this was a "what's left" situation)
• Soy sauce (the everyday bottle, not anything fancy)
• Lemon (one, slightly past its best)
• Jasmine rice (half a bag)
• Mixed spices: cumin, smoked paprika, dried oregano, a bit of chilli flakes

This is a deliberately undramatic list. It's not a stocked chef's pantry. It's a Tuesday night fridge with a few pantry defaults behind it. The interesting question wasn't "can AI cook with these" — of course it can. The question was what each cuisine would do with them.

---

Vietnamese: Lemongrass Chicken Rice Bowl

The Vietnamese result came back as a ginger-lemongrass chicken rice bowl with a quick nuoc cham-style dressing.

What's interesting here is what the AI added from pantry staples — and more importantly, what it didn't change. The soy sauce stayed but got used sparingly, as a background note of umami rather than the primary flavour. The lemon became a lime stand-in for the dressing, squeezed over at the end rather than cooked in. The garlic was sliced thin and briefly fried to a crisp, scattered on top as a textural element. The chicken got poached rather than fried or grilled, pulled apart while still warm.

Vietnamese cooking tends toward freshness and brightness — the acid is assertive, the herbs are abundant, the heat comes from chilli rather than spice blends. In this recipe, the cumin and paprika didn't appear at all. The AI correctly understood that those spices would read as wrong in this context. Instead, lemongrass and ginger were brought in as pantry-staple assumptions (fair enough — they're common enough to presume), and the dish ended up clean and fragrant where I'd started with something quite heavy-handed and earthy.

The macros came back at around 480 calories, 36g protein. Reasonable for a bowl meal.

---

Lebanese: Garlic Shawarma Plate with Lemon Toum

The Lebanese result was a garlic-spiced chicken shawarma plate — charred thighs over rice, with a simplified toum (the Lebanese garlic sauce) made from the lemon and most of the garlic.

This one was the biggest transformation. The soy sauce disappeared entirely. The cumin and paprika came to the front, supplemented with a prompt assumption of cinnamon and allspice — spices that are foundational to Lebanese cooking but that I didn't mention. The garlic stopped being a background aromatic and became the headline: roasted in the pan drippings, then blended into the sauce with lemon juice and a splash of water.

The chicken thighs got treated completely differently here — scored, spice-rubbed, and cooked at high heat to get some char on the skin. Lebanese cooking is about that contrast between the slightly bitter char and the bright, acidic sauce against something fatty. The dish felt substantial in a way the Vietnamese version didn't. Same protein, same rice, genuinely different experience.

Macros: 520 calories, 40g protein. The toum added a bit of olive oil, which the AI correctly folded in as a cooking staple.

---

Korean: Gochujang Chicken Rice Bowl

The Korean result came back as a gochujang-glazed chicken rice bowl with sesame and spring onion.

Here the soy sauce came back to the centre — used in the glaze alongside the assumed gochujang (the fermented chilli paste that's the backbone of a huge proportion of Korean cooking). The garlic got treated with the most aggression of the three cuisines: minced fine, almost paste-like, used in quantity and cooked at higher heat. The lemon made a brief appearance as a finishing squeeze but wasn't a structural part of the dish the way it was in the Vietnamese version.

What struck me was how the rice was treated. In the Vietnamese bowl, the rice was steamed and served as a neutral base. In the Lebanese plate, it was cooked in stock with a little spice. In the Korean version, the rice became a platform for sauce: the gochujang glaze pooled at the bottom of the bowl and got mixed in by the eater. The dish is built to be actively combined, not plated neatly. There's an interactivity to Korean bowl food that the other two cuisines don't quite have in the same way.

Macros: 510 calories, 38g protein. The sesame oil pushed the fat up slightly.

---

What This Actually Shows

Cuisine isn't really about ingredients. It's about frameworks — the combination of technique, spice logic, acid placement, and cultural expectation that determines what a meal is for and how it should feel to eat it.

The same chicken thighs can be pulled and dressed (Vietnamese), charred and sauced (Lebanese), or glazed and combined (Korean). The same garlic can be barely-there and crispy, the dominant flavour note in a sauce, or aggressively caramelised into a glaze. The same lemon can be an assertive finishing acid, the base of an emulsified sauce, or almost invisible. None of those transformations require different ingredients — they require a different understanding of what you're building.

This is why I find the Cuisine Explorer genuinely interesting. I don't use it because I'm bored of my usual cooking — I use it because I want different frames for the same starting point. When I scan my fridge and then pick a cuisine, I'm not asking "what can I make with these ingredients?" I'm asking "what would someone who grew up cooking Vietnamese food do with these ingredients?" Those are different questions with different answers.

Seventeen cuisines in the app. The same pantry every time. Genuinely different results each time.

---

If you want to try this at home: pull open Cuisine Explorer, pick a cuisine you don't cook often, and see what it does with what you've got. The gap between "I only have the basics" and "I could make something interesting tonight" is usually narrower than it looks.