General-purpose AI assistants are not built specifically for cosmetic formulation, so their formulas often lack critical details such as phases, processing temperatures and preservative strategy, and may contain errors that a trained cosmetic chemist would spot immediately. When you ask ChatGPT to formulate a moisturizing serum, it will produce plausible-sounding output: glycerin, hyaluronic acid, humectants, and an emollient oil, often with percentages that fall within normal ranges. However, the response typically omits or oversimplifies several things that matter in production. First, it may not specify which grade or molecular weight of hyaluronic acid (1.8 MDa vs. 50 kDa behave very differently in a formula and affect viscosity and feel), may not call out the pH sensitivity of your chosen actives, may forget to include a chelating agent if your water source is mineral-rich, or may suggest a preservative system that is legal in North America but restricted in the EU. Second, general AI does not know your manufacturing partner's capabilities: if your co-packer uses a single-vessel mixer and no heating, formulas requiring phase inversion or high-temperature emulsification will fail. Third, it has no cost data. A formulation that looks perfect on paper may use a $400/kg active in a 5% load, making the raw material cost $20 per liter before packaging, labor, or profit—far above your target. Fourth, general AI cannot run stability predictions or advise on the testing protocols you should follow: it may not warn you that a oil-soluble active like vitamin A palmitate is prone to isomerization under heat and light, or that a high-pH cleanser will degrade niacinamide. Finally, general AI has no memory of what worked or failed in similar formulas, so it cannot advise on the hidden gotchas—for example, that adding more than 3% glycerin to an oil-based product can cause the emulsion to break, or that certain essential oils are incompatible with certain preservative systems. The result is that founders often spend weeks testing a general-AI-generated formula, discover multiple problems (instability, poor sensory feel, regulatory non-compliance, or cost overruns), and then hire a real chemist to rebuild it—which is more expensive and slower than starting with a human chemist or using a chemist-vetted AI tool from the beginning.
Ingredients in this answer: Hyaluronic Acid · Glycerin · Niacinamide
Read the full guide on Is There an AI Tool That Can Create Cosmetic Formulations? →