Just for fun, we asked ChatGPT to suggest ten ice cream flavour combinations that would work well together but which we haven’t tried before.
The result was partly predictable (Earl Grey and lavender sounds lovely, but we’re sure it’s been made already), fairly interesting (black sesame coconut is right up our street) and sometimes downright strange (black garlic chocolate chip, anyone?).
But beyond our trivial experiment, what’s the future of AI in food development?
Following our blog post on the pros and cons of using AI to translate your product label, we’ve looked at how artificial intelligence is changing how food and beverage products are created and reformulated.
AI can accelerate the product development process
The big difference between AI tools and human-led product development is the speed of data capture and analysis.
Gone are the days of trial and error; AI can collate and process vast lakes of information from multiple sources in seconds, converting it into valuable insights. It can even factor opinions and emotions into its data calculations.
Rapid number crunching is a helpful way to get ahead of consumer trends. For example, Olam Food Ingredients (OFI) recently used AI-powered research to scan online recipes, restaurant menus and food websites, tracking flavour preferences.
OFI found interest in Asian flavours surging in the West, with dragon fruit and sesame gaining popularity in Europe and lychee and yuzu getting traction in the USA. AI analysis also discovered that Western flavours like butterscotch, marshmallow, and salted caramel were taking off in India and Indonesia.
UK supermarket chain Waitrose also recently used AI insights to bring 65 new products to market, revealing a consumer appetite for US-inspired dishes like hot honey pork belly slices, mac' n cheese quiche and Tex Mex creamed corn dip.
An evolution, not a revolution?
AI’s data analytics capabilities are quick and powerful, but OFI’s Edward Norder warns that food and beverage companies “need to be able to pair the right flavours and translate emerging trends into great-tasting products in the development kitchen.”
Global consumer food strategist Cyrille Filott agrees with this sentiment, stating that AI’s influence is evolutionary rather than revolutionary.
“You could take an existing product and you could tweak perhaps one of the ingredients, like the flavour… but I think developing something completely new is still a long way away because you need to change lines and you need to create new packaging formats and all that,” he tells Just Food.
There are already examples of food brands using AI to refine existing products or expand a successful product range. For instance, Unilever used AI to analyse millions of flavour combinations in developing its Knorr zero salt veggie stock cubes.
Unilever also used AI silicon modelling to predict the taste, texture and stability of ingredients when creating Hellman’s vegan mayonnaise. Ultimately, they replaced the egg with modified corn starch to retain the same creamy texture as traditional mayo in its plant-based offering.
For the most ambitious companies, AI could become a valuable tool in the development of novel foodstuffs such as proteins and enzymes, accelerating the sustainable food movement.
AI can’t replace human creativity – yet
While AI can steer food companies towards intelligent decisions, we’ve not reached the point where all new products can be developed through automation.
AI can identify trends and solve specific formulation issues, but food and drink is a creative industry. Sometimes, products that should make sense statistically just don’t resonate with consumers.
One example is Coca-Cola’s Y3000, an experimental drink developed through AI analysis of a major tastes and trends survey. The limited edition flavour received a resounding thumbs down upon release; T3 journalist Rik Henderson declared, “if this is the future, I'm happy where I am”.
There’s also consumer readiness to consider. While YouTuber RachhLovesLife has great fun testing different recipes recommended by ChatGPT, we’re doubtful people will rush to recreate her feta ice cream!
NPD and reformulation strike a delicate balance between pushing boundaries and providing comfort, some of which is instinctive.
The rest is built on an innate knowledge of consumer tastes – globally and regionally. A flavour profile that works well in one market won’t always resonate in other parts of the world (for taste and cultural reasons).
For example, our blog on NPD and compliance discusses how Coca-Cola sweetens its products with cane sugar rather than corn syrup in Mexico because local communities are passionate about preserving the country’s sugar trade.
Developing new products that consumers WANT to buy
The food industry has only touched the tip of the AI iceberg, but we can already see a huge opportunity for brands to better understand consumer behaviours and preferences.
The challenge is converting this knowledge into an end product people want. As the saying goes, “knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad.”
The world is unlikely to be ready for black garlic chocolate chip ice cream anytime soon, but AI could help us pioneer novel foods and new techniques that support quicker, more successful NPD.
For food and drink brands, the key will be using AI insights and capabilities in conjunction with the skills and knowledge of people who understand regional behaviours and how they could impact your next recipe.
Hooley Brown helps brands localise product formulas, packaging and labelling for multiple markets.
Our global network of experts can help you understand the needs of consumers in each region and develop products that meet their preferences while complying with local legislation.
Get in touch to find out more.
