- Packaging Design
According to OpenAI, in September 2022 over 2 million images a day were created through its DALL-E 2 platform. This number is only likely to have increased as the public become more aware of generative AI. Everything from logos and avatars to competition-winning artworks are being generated by AI. Unsurprisingly, designers are also experimenting with creating packaging concepts. Last year, Ludovic Mornand of Studio Blackthorns ran an experiment to see whether AI could create a viable packaging design. Mornand used two AI tools to create packaging illustrations (Hypnogram.xyz) and copy (Rytr) for a non-alcoholic ready-to-drink concept brand called Alt Planets. According to Mornand, the final packaging design was 80% the work of AI and 20% human. Silas Amos described running a similar experiment designing a bottle for a theoretical Kylie Minogue fragrance in Dieline. While we don’t believe that AI image generators are currently capable of handling every aspect of packaging design, the results of experiments like these mean it is likely that brands will use them to help generate initial concepts.
- Non-Visual Design
Incredibly, AI is also playing a helping hand in non-visual design and branding. Scent, taste and touch are particularly difficult senses to translate digitally. Yet, Givaudan - the world’s biggest fragrance and flavour manufacturer - is using AI as a fragrance co-creation tool. ‘Creatogether’ is a collaborative solution from Givaudan and Tmall that helps brands to create new fragrances with AI. This includes everything from creative ideation to instant sampling. Givaudan is also using AI to help consumers buy perfume online. Its Myrissi algorithm translates scents into colour patterns, based on data from over 25,000 consumer tests. These trigger the same emotions in customers as the perfume scent, helping potential customers to make informed buying decisions when they can’t smell the fragrance. The technology also helps to direct packaging design by providing suggested colour palettes that match the scent of the perfume. This could open up a new aspect of branding where colours and design are based on AI interpretations of the themes, emotions and feelings that a brand wants to be associated with.
One of the hardest parts of branding is creating packaging that translates well from 3D to a small, flat image online. While designers can work to make sure they consider both the physical and digital channels, they don’t necessarily have a way of measuring whether they got the balance right. AI-powered tools can help. Vizit, described as the world’s first visual intelligence company, uses AI in its image analysis platform to help brands optimise their designs. The company has worked with Mars over the past two years with great success. The AI can help brands to recognise issues with an image at an earlier stage in order to refine the design. This includes thinking about the potential success of the design at an individual retailer level or for a specific audience. According to an interview with Packaging Digest, Vizit’s enterprise customers can get insights 99% faster when pretesting packaging design concepts by using AI and at a 10x cost reduction. One particular advantage of using AI as a ‘final check’ of a design is that it can help reduce biases from individual designers or teams. Even if you don’t agree with its assessment or find it controversial, it may provide valuable insight if you step back and look at why you don’t agree. We expect to see more and more brands using AI tools to optimise the packaging designs created by their human teams to ensure they perform as well as possible.
Sustainability is high on the agenda for many brands. AI is one tool that can help with this. Since 2018, Firmenich, the Swiss fragrance company, has provided an AI-powered tool for measuring fragrance sustainability. The EcoScent Compass helps the company’s perfumers to develop more sustainable fragrances. It can measure and calculate a large number of data points for each ingredient, including the impact on people, climate and nature, and circularity. In turn, the system makes it easy to communicate the sustainability credentials of products to suppliers, partners and consumers. This could include sustainability information for packaging. Packaging end-of-life is another big focus area for AI. AMP Robotics uses an AI platform and robotic arms to pick out recyclable materials from waste. The system can identify different colours and shapes of packaging as well as logos and branding. Last year, a new consortium launched with the aim of using AI to improve packaging waste sorting. Members include major brands - Colgate-Palmolive, Danone, Ferrero, LVMH Recherche, Mars, Michelin, Nestlé, PepsiCo, Procter & Gamble - as well as Ghent University, Radboud University, and the National Test Centre Circular Plastics in the Netherlands. Over the course of two years, the consortium will test product packaging from the partner brands to develop an advanced AI model for sorting packaging. This includes separating food and non-food packaging. The aim is to introduce the AI developed to sorting plants in Europe at the end of the project. Not only can AI help brands to identify, sort and separate recyclable packaging in order to adopt a circular model, but it may also provide valuable data that could inform packaging design. For example, making recycling symbols more prominent or adopting certain colour schemes that can more easily be recognised at sorting facilities.
- Brand Recognition
If you’re curious about how recognisable your branding is, why not try asking AI image generators to draw the product you make and see what they come up with? AI chatbot technology is trained on datasets, such as text and images, in order to generate responses to prompts. In the case of images, this may include art that can be identified as belonging to a specific style. Last year, Heinz asked OpenAI's DALL-E 2 to draw ketchup. The responses all resembled a Heinz ketchup bottle and label (at least according to the results shared by Heinz). It follows a campaign in 2021 where Heinz asked people around the world to draw ketchup. Similarly to the AI, the drawings all feature recognisable Heinz branding which suggests a strong association between the product (ketchup) and the brand (Heinz). Ok, so AI image generators aren't really a meaningful way of measuring brand recognition. For Heinz it makes for a nice marketing campaign, but even then we are assuming they filtered out any results that didn’t look like their branding. But they might be a fun way to help inform your understanding of what features come up again and again for your target product or category. The results generated by these AI chatbots are based on patterns and predictions of what elements go together. And this ‘thinking’ is informed by thousands of examples within the training dataset. This means you can essentially see the crunching of all of that data in the results your prompt generates. So if you search for ‘cereal’ or ‘milk’ or whatever, you’ll get a strong sense of what colours and shapes are associated with that product very quickly. You may then be able to use this to help inform your branding for stronger consumer recognition. Or if you’re really lucky, like Heinz, you might find that your branding and product association are so strong that you’re looking at your own product in the results.
By Jack Stratten, Head of Trends at retail trends consultancy Insider Trends.