Personalized nutrition platform Qina unveils tips for ethical AI tech

25 Jan 2024 --- Aiming to provide an operational AI framework for companies, practitioners and developers in the nutrition, F&B, digital health and ingredients sectors, strategic consultancy Qina shares its insights on AI use in a newly released white paper — “The ethics of artificial intelligence (AI) at the intersection of nutrition and behavior change.”

“By collectively raising awareness of the risks and opportunities associated with AI systems in nutrition and behavior change, we can uphold and protect our societal values and morals in an ever-growing world of machines,” says Mariëtte Abrahams, CEO and founder of Qina.

“We are proud to provide leadership on this important topic, which will have societal implications for years to come,” she continues. “While we are a digital-first company incorporating AI in many areas of our own business, we feel it is our role to raise the alarm when it comes to nutrition and health. We hope that with this white paper, we can shed some light on an important area and spark some focused internal discussions about potential and existing blindspots.” 

Opportunities in personalized solutions
The main pillars of the Qina AI ethics framework center around nutrition, technology, health and society and include considerations for the importance of a human-centric approach, organizational specificities, education and training, people and planet, data and regulation. 

Among the key messages communicated in the paper is the assertion that ethical AI solutions require human oversight and collaboration between stakeholders in the nutrition industry and that the availability of ethical and trustworthy solutions is vital to the building of trust in the new technology. 

The spotlighted opportunities of AI use in nutrition refer to its potential to improve access and affordability of personalized solutions and it can identify hidden patterns in disparate datasets. 

Recent research by the Taipei Medical University, Taiwan, revealed that freely available AI chatbots are able to provide nutritional estimates comparable to those of nutritionists in training, which one of the study authors outlined as a win for the accessibility of personalized nutrition. 

Some recent innovations in the field of AI personal nutrition include a solution to malnutrition in clinical settings, meal sustainability optimization and genetically personalized nutrition-based grocery shopping. 

AI limitations
The pertinent challenges highlighted in the document is that the current training of AI with datasets that are unrepresentative of society, which could lead to an increase in inequality, the lack of transparency and strong regulation. 

The inherent bias and lack of ethnic and gender diversity of datasets fed to AI should be taken into consideration in the creation of AI personalized nutrition solutions.The origins of datasets fed to AI are a popular point of discussion in the current discourse on AI ethics due to their inherent bias, the lack of ethnic and gender diversity. This inherently relates to nutrition. 

One example is genetic databases, built overwhelmingly on research conducted on people of North American and North European descent. This can lead to dietary and lifestyle recommendations being irrelevant, discriminatory or inaccurate for people of other ethnic backgrounds.

For instance, people from West African regions are more salt-sensitive than Europeans, which is a cardiovascular disease factor likely to go unnoticed by AI at present.

Similarly, AI algorithms may omit or misrepresent social determinants of health (SDOH), the conditions in which people are born, grow, live, work and age, and the effects these factors have on health. The whitepaper recommends the integration of SDOH into AI personalized nutrition for equitable health outcomes.

 AI nutrition could threaten food cultures by favoring popular foods consumed by the masses in wealthy nations, particularly those with high disposable incomes. This could potentially influence the agri-food sector by dictating consumer trends. 

Thus, the solutions Qina outlines concern the need for education in order to increase agility and efficiency in developing AI solutions that are patient-centered. It points to the social responsibility that dictates data is shared among stakeholders. 

The proposed framework acknowledges the importance of existing AI regulations on ethics, such as the European Commission’s 2019 Ethics Guidelines for Trustworthy AI of the High-Level Expert Group, the 2019 OECD AI Principles or the 2020 Harvard University, US, Principled AI framework, which the company describes as being part of the collective effort to ensure AI safety. 

However, it also states that these documents do not go far enough in acknowledging the importance and high impact of nutritional science. 

By Milana Nikolova