We see Artificial Intelligence (AI) cropping up in more and more industries. It prompts suppliers, brands and even Cosmetic Chemists to ask: is there a place for AI in cosmetic formulation?

In this article, we’ll explore exactly what AI can do for cosmetic formulation, and where it simply can’t replace the skills and knowledge of a qualified Cosmetic Chemist.

AI is a great tool for preliminary research.

AI can be used effectively when running preliminary searches into consumer attitudes, cosmetic ingredients, and product types. This can give anyone conducting the search some insight into what information currently exists regarding consumer preferences as well as what the internet has to say, from various sources, about cosmetic ingredients. It can also enable Cosmetic Chemists and even Brand Managers to quickly identify which cosmetic ingredients are trending and most commonly sought in specific product formulation types. Watch this video to learn how to spot trends without the use of AI.

AI’s abilities can be limited beyond preliminary research because:

• it cannot replace a proper regulatory search on limitations on inputs of various ingredients to ensure safety and compliance;
• the results don’t provide trade names which are essential when sourcing particular grades or concentrations of ingredients;
• it’s search results cannot be used as evidence to support the input or efficacy of a cosmetic ingredient;
• listing the ingredients commonly found in a certain type of cosmetic formulation does not identify where innovation can give a cosmetic brand a strong and marketable point of difference.

It takes a properly trained Cosmetic Chemist or Formulator to accurately check regulatory limits, determine required inputs for efficacy, evidence-based claims, stability and quality, and most importantly, incorporate a particular cosmetic brand’s specific philosophy and formulation principles.

AI results may be skewed by popular misinformation

AI relies on the information that exists on the internet to provide its search results. This means that the information provided is based on what is most commonly written or referred to. It currently does not have a way of knowing whether it is providing the correct answer, but instead, the most common responses.

This means misinformation may be transmitted as an AI response. As an example, there are multiple cosmetic ingredients with popular yet misinformed safety profiles on the internet; and AI will repeat this information without knowing it is not factually correct.

There may also be fear-mongering campaigns or other incorrect information about the use of certain ingredients in cosmetics which is not accurate. Again, where such information is widely spread throughout the internet, AI can only report on what it finds.

This means search results using AI about cosmetic ingredients may be more than just a little inaccurate. It can give a cosmetic Brand Manager or Formulator information on how consumers may view a cosmetic ingredient, but it doesn’t give the full truth, correct regulatory status or even use of an ingredient where such misinformation exists. Try googling parabens in cosmetics and you’ll soon see the inaccuracies of the information provided!

AI cannot yield a cosmetic formula with specific aesthetic or performance characteristics

While AI can be used to give a ‘typical’ starting guide to how much emulsifier might be used in a cream or lotion, along with common inputs of lipids and preservatives, it cannot give you precise information on exactly which emulsifiers, lipids or preservatives - especially with trade names - should be used.

Other cosmetic formulation AI limitations include:

• it cannot predict the sensory and aesthetic aspects of ingredient combinations like an experienced Cosmetic Chemist can;
• it cannot assess and adjust cosmetic formulation inputs to meet specific target market needs or preferences;
• it’s approach to natural and sustainable cosmetic ingredient choices;
• it’s ability to create multi-ingredient formulations to fulfill the needs of specific product development briefs;
• being certain of safety, regulatory and stability outcomes of a cosmetic formula;
• manipulating inputs of a cosmetic formula to address sensory or aesthetic needs of a product;

making innovative cosmetic ingredient choices, because it relies on the information of what already exists.

What could AI do for cosmetic formulation?

Imagine if AI could accurately:

• use details of an existing formula and hue to accurately predict how to create alternate shades using specific colour additives;
• run multi-country cosmetic ingredient compliance checks;
• be used to run a machine that could create minor but important multi-sample variations for hastier lab developments;
• predict SPF outcomes in specific o/w or w/o emulsions, including the impact of any required formulation enhancements or variations on aesthetics and stability;
• predict consumer preferences without needing to run the tests with consumer groups.

The first four of the above items take highly trained Cosmetic Chemists significant time at the bench whilst creating various product samples, and the final evaluations take Brand Managers considerable time with consumer testing. Imagine if AI could short cut some of the guess work - items it can’t currently accurately complete because it cannot predict unknown outcomes. This is where the human approach to cosmetic formulation and product evaluation cannot currently be replaced.

What is the best place for AI in cosmetic formulation?

At this point in time, AI should not be relied upon for specific cosmetic ingredient inputs, especially when what cosmetic brands really want are innovative, efficacious and competitive formulations to suit their specific target markets. AI can however serve as a good starting point for market research, to understand consumer perceptions and which cosmetic ingredients are commonly used in different cosmetic formulation types.

AI cannot be used to make decisions but instead guide trained Cosmetic Chemists and Brand Managers in what they may need to investigate further - and what approach they may need to take to create true innovation.

We may, in the near future, see AI incorporate better predictive technologies or even harness regulatory databases in an accurate and accessible way. These would be potential applications with real benefit to Cosmetic Chemists, who can then focus on the sensory, stability and evidence-based results while creating innovative and effective cosmetic products of the future.

Happy formulating!