7.2 AI & Machine Learning in Aroma Compound Selection

Artificial Intelligence (AI) and Machine Learning (ML) are revolutionizing the perfume industry by helping perfumers discover new aroma compounds, optimize fragrance formulations, and predict scent interactions more efficiently. These technologies reduce trial-and-error experimentation, making fragrance development faster, more cost-effective, and sustainable.

1️⃣ How AI & Machine Learning Work in Perfumery

Traditional fragrance creation relies on expert perfumers who manually test and blend hundreds of ingredients. AI and ML, however, analyze vast datasets of aroma compounds, predict scent combinations, and suggest optimal formulations using pattern recognition and algorithms.

Machine learning models learn from:
✔ Aroma databases – Thousands of known fragrance molecules and their properties.
✔ Human feedback – Consumer preferences and olfactory perception data.
✔ Chemical structures – Predicting how new molecules will smell based on their molecular structure.
✔ Environmental & stability data – Ensuring long-lasting and safe formulations.

💡 Example: Givaudan, one of the world’s largest fragrance companies, uses AI-powered software called Carto to suggest fragrance formulations based on previously successful scents and molecular data.

2️⃣ Benefits of AI in Aroma Compound Selection

✅ Faster Innovation: AI can suggest new fragrance formulas in seconds, compared to weeks of human testing.
✅ Cost-Efficiency: Reduces waste and unnecessary trials by predicting successful blends early.
✅ Sustainability: AI helps identify eco-friendly and biodegradable alternatives to traditional aroma chemicals.
✅ Enhanced Creativity: AI augments perfumers’ creativity by suggesting unexpected but harmonious ingredient combinations.

💡 Example: The fragrance brand Philyra (IBM AI) helps design personalized perfumes by analyzing a person’s preferences and body chemistry.

3️⃣ AI in Action: How AI Selects Aroma Compounds

🔹 1. Predicting Scent Profiles of New Molecules

AI analyzes the molecular structure of a chemical and predicts whether it will smell floral, woody, citrusy, or musky before it is even synthesized.

✔ Example: Researchers used Deep Neural Networks (DNNs) to predict the scent of new molecules without needing human sensory testing.

🔹 2. Replacing Allergenic or Banned Ingredients

Regulations often restrict certain ingredients due to health concerns or sustainability issues. AI suggests safe alternatives that mimic the same scent.

✔ Example: AI helped replace Lilial, a floral compound banned in the EU, with a similar-smelling but non-allergenic molecule.

🔹 3. Creating Long-Lasting Fragrances

AI predicts the evaporation rate of different molecules and suggests fixatives that make perfumes last longer on the skin.

✔ Example: AI recommended Ambrettolide (a biodegradable musk) as a sustainable alternative to Galaxolide (a non-biodegradable musk).

🔹 4. Personalizing Perfumes Based on Consumer Preferences

AI-powered tools analyze consumer feedback, purchase history, and scent preferences to suggest tailored perfume formulations.

✔ Example: Firmenich’s AI-driven fragrance tool customizes scents based on individual preferences.

4️⃣ Practical Example: Using AI to Select Aroma Compounds for a New Perfume

Let’s create a modern AI-assisted fragrance formula using an AI tool to select optimal aroma compounds.

🔹 AI-Suggested Perfume Formula (Fresh Floral Musk for Women)

AI Inputs:

• Target audience: Women, aged 25-40
• Preferred scent family: Fresh floral musk
• Sustainability requirement: Biodegradable, cruelty-free, and eco-friendly

AI-Recommended Formula:

Top Notes (Fresh & Fruity – 40%)
✔ AI-selected molecule: Limonene (from citrus peels) – 10%
✔ AI-selected molecule: Calone (marine/aquatic note) – 10%
✔ AI-selected molecule: Green apple aldehyde – 10%
✔ AI-selected molecule: Pear ester – 10%

Middle Notes (Floral & Sweet – 30%)
✔ AI-selected molecule: Jasmine lactone – 15%
✔ AI-selected molecule: Rose oxide (green floral note) – 10%
✔ AI-selected molecule: Helional (airy floral) – 5%

Base Notes (Musky & Long-Lasting – 30%)
✔ AI-selected molecule: Ambrettolide (biodegradable musk) – 10%
✔ AI-selected molecule: Cetalox (ambergris replacement) – 10%
✔ AI-selected molecule: Amyris oil (sustainable sandalwood alternative) – 10%

✔ Solvent: AI-selected sugarcane-derived ethanol – 80mL (80%)

✅ Final Product: A 100mL AI-designed Eau de Parfum with an optimized balance of longevity, freshness, and eco-friendly ingredients.

5️⃣ The Future of AI in Perfumery

🔹 AI-generated fragrance molecules: AI will design entirely new aroma compounds that do not exist in nature.
🔹 Fully personalized perfumes: AI will create fragrances based on a person’s DNA and skin chemistry.
🔹 Augmented reality scent selection: AI-powered AR tools will allow consumers to “smell” a fragrance digitally before buying it.
🔹 AI-powered sustainability tracking: AI will help brands ensure carbon-neutral and eco-friendly formulations.

6️⃣ Final Thoughts & Summary

🚀 AI & Machine Learning are revolutionizing aroma compound selection by:
✔ Predicting scent profiles based on molecular structure.
✔ Replacing banned or allergenic ingredients with safer alternatives.
✔ Creating long-lasting perfumes with optimized fixatives.
✔ Personalizing fragrance formulations based on consumer preferences.

🌱 Next Challenge: Try an AI-based fragrance tool (like IBM Philyra or Givaudan Carto) to create a personalized perfume formula!