Findings of the survey were published on the journal Food Supplements and Biomaterials for Health (FSBH) published by Health Functional Food Future Forum.
The research was done by Korea University Business School’s Department of Marketing, and Dr Sun Ho Kim, president of Seoul-based health functional foods (HFF), cosmetics and food regulatory consultancy firm SEAH Bio solution.
Dr Kim is also the developer of the AI method evaluated in the survey.
Three hundred participants with an average age of 38.5 years old took part in the online survey conducted on Amazon Mechanical Turk.
They were randomly divided into two groups of 150 each.
The two groups were told to either read 1) the process behind manufacturing a new HFF using AI or 2) the conventional method, which relied on humans in selecting and formulating ingredients for a new HFF.
The AI model assessed in the survey was NaturaPredicta™.
It applies natural language processing (NLP) to the PubMed database to predict the bioactivity and similarities of various botanical ingredients.
Compared to the conventional method, it is said to significantly reduce the time and cost required to identify ingredients for developing new HFF.
The participants were then asked to rate HFF developed using AI and the conventional method based on four criteria – trustworthiness, expertise, attitude towards the products, and purchase intention.
Questions asked include “I can see the benefits of an AI prediction model that can perform ingredient identification better than humans”, “I think the AI prediction model is competent”, “I think this HFF is more nutritious than average”, and “I am going to buy this HFF”.
The participants then gave their answers on a scale of one to seven. The higher the points, the higher they agree with the questions and statements listed in the survey.
Significantly higher ratings for the AI-method
Results showed that participants gave significantly higher ratings for the AI-method, as compared to the conventional human-method on all four fronts.
For example, trustworthiness towards HFF developed using AI score was 5.29 ± 1.08, which was 4.33 ± 1.36 for conventional method.
Perceived expertise of the AI model was also higher at 6.12 ± 0.91 versus 5.01 ± 0.92.
The greater trust and perceived expertise of the AI model also led to a more positive attitude of products developed using AI. In this case, products developed using AI scored 5.91 ± 1.19, while the conventional method scored 4.65 ± 1.03.
Lastly, the respondents were more likely to purchase HFF developed using AI – with a rating of 5.17 ± 1.17 as compared to the conventional method at 4.24 ± 1.28.
“The results indicate that consumers perceive an AI system as trustworthy and highly expert, leading to favourable attitudes toward the products and increased purchase intentions,” said the researchers.
They added that using AI could also cut the amount of time and incidence of errors when formulating a new HFF.
“Traditionally, developing a new HFF has heavily relied on human expertise and manual processes for classifying, predicting, and identifying the raw materials or ingredients needed.
“This method can be prone to variability and errors. However, AI technologies help mitigate these issues by providing consistent and reproducible predictions, thanks to their ability to handle large data sets with remarkable accuracy,” the researchers added.
However, they also cautioned that since the participants were relatively young, the results could be biased and not reflective of the general Korean population.
They urged future studies to aim for a more representative sample by including participants of varying ages, from different countries, and with diverse cultural backgrounds.
Companies using AI for NPD
New Zealand’s SRW Laboratories is an example that is using AI to develop new longevity nutraceuticals.
It is working with AI drug discovery firm Insilico Medicine in this endeavour.
The partnership will first focus on the next-generation anti-ageing and longevity products.
China’s dairy giant Yili also developed its internal version of AI – dubbed Yili-GPT – for product innovation.
However, the company raised one caveat – that AI was useful but limited.
It added that while AI could speed up the innovation process, it was not expecting the technology to replace human ingenuity.
Source: Food Supplements and Biomaterials for Health
Effect of AI-Driven Methods for Health Functional Foods on Consumer Acceptance: Evaluating Trust and Expertise Perception
https://doi.org/10.52361/fsbh.2024.4.e14
Authors: Jeongbin Whang and Sun Ho Kim
