The role of AI in Food and nutrition innovation

Oct 18, 2024 11:21am

 

Artificial Intelligence technologies such as Generative AI are set to disrupt industries, the way we work and employment for the coming years. While there is a lot of excitement and hype, there are even more fears of what this rapid shift will mean for innovation and society. In this article we share our insight and perspective on how AI is currently leveraged in the context of Food & Beverage, nutrition and Ingredient industries as well as how the new regulation, the EU AI Act will impact these verticals.

 

Written by: Mariëtte Abrahams PhD MBA - Founder & CEO Of Qina

 

Introduction to the AI Act

The EU AI Act came into force on the 1st of August 2024. This new regulation and legal framework aims to provide oversight and accountability for the development and use of artificial intelligence (AI) in products and services. The purpose is to ensure human-centric and trustworthy artificial intelligence (AI), strengthen uptake, investment and innovation in AI across the EU. The AI Act applies to providers and deployers of AI systems within the EU, as well as those in third countries if the AI system's output is used in the EU.

 

The EU AI Act came about after years of consultation following concerns around the rapid advances in AI technologies, and how this could impact individuals and wider societies. These concerns have been replicated in numerous scientific publications (2) and public stakeholder groups, as well as being extensively covered in the media. 

The AI Act divides the implementation of AI according to different risk levels. Risk is, according to the Act, the combination of the probability of an occurrence of harm and the severity of that harm.  

The risk levels include:

  • Unacceptable, such as social scoring
  • High
  • Limited 
  • Minimal 

General-purpose AI such as chatbots using ChatGPT are also included under the AI Act.

The majority of concerns around the employment of AI, are in the areas of education (3), recruitment (4) and law enforcement (5), which have directly led to individuals being excluded, ignored, and unlawfully arrested. Whilst this sounds extreme, companies employing AI in the Food & Beverage industry are not immune from this new regulation. 

This article outlines how AI is currently applied in the Food & Beverage industry and what executives in R & D, innovation and product management need to have in place in order to comply with the new regulation.

 

The why of AI in Food and Beverage?

AI is a powerful technology which has advanced at an unprecedented pace (6). In a world where the global population is growing and aging, where 20% of food produced is wasted in homes (7), where our climate is rapidly changing and the prevalence of lifestyle-related chronic diseases is exploding, AI can be a key lever in changing the course. With consumer interest in consuming healthy products at an all-time high, AI can be used to increase access to nutrition and health information. In combination with new technologies such as generative AI, there’s an opportunity to open up novel ways to feed, nourish and inform consumers at the intersection of food and health.

 

20% of food produced is wasted in homes

 

How is AI currently applied in Food & Beverage? 

AI has already been employed in the Food & beverage industry for years,with some use cases being consumer-facing and others for in-house operations.

Here are a few examples of how AI is currently being leveraged in the Food & Beverage and Ingredients industry:

  • Discovery of bioactive compounds: AI is used to analyze vast datasets from scientific literature and food composition databases to identify bioactive compounds in upcycled foods that have health properties. These compounds can be included in products to enhance their health or nutritional profile. For example, Nuritas has discovered functional peptides by leveraging AI. Their patented ingredients have been demonstrated in clinical studies to mitigate muscle atrophy, reduce signs of aging and reduce levels of HbA1C (8)
  • Meal, recipe and food shopping recommendations: AI and machine learning can be leveraged to combine disparate data sets in order to develop personalised recommendations. For example, EatLove is a smarteating tool that uses personal data, such as taste and dietary preferences, as well as a health profile, to provide personalized recommendations. AI is leveraged to match and combine the personal profile to food ingredients and product SKU’s in store. Individuals receive meal plans and shopping lists tailored to their needs in seconds. Their chatbot “Ava” uses generative AI to provide support and feedback.
  • Food labelling: AI and machine learning are used to analyze, tag and enhance nutrition information for transparency. For example, Nutritics can analyze menus, product information and recipes to provide nutritional analysis based on nutrition and health claims.
  • Food as medicine: (Spoonguru)- AI can also be used as part of the “Food as Medicine” movement to develop “food scripts” or guide individuals to ingredients or products that can contribute towards their health as part of their dietary program. For example, Spoonguru leverages AI by combining datasets such as clinical guidelines, food composition, and in-store products to help consumers navigate the store shelves with products that match their health needs.
  • Behaviour change: (Zoe)- AI is also increasingly used to nudge individuals towards products that closely match their personal preferences, beliefs and health needs, by using a single score. In this instance, AI combines disparate datasets that could include personal, biological, product, and sustainability data (e.g. carbon footprint). A user would then scan a product using the barcode scanner feature on their smartphone and see the information presented as a single score in an app, allowing them to learn how well the product matches their needs and goals. For example, Zoe uses personal and biological data to develop a meal score enabling users to see how well products they plan to consume match their profile and preferences. 

 

What are the ethical challenges of using AI?

Algorithmic bias

There are important challenges with immense consequences that need to be considered and corrected when employing AI which include: algorithmic bias, a lack of transparency of AI systems, the use of unrepresentative datasets to train AI systems, and the risk of widening inequality when AI development and datasets primarily come from digitally literate groups with greater access to technology, potentially skewing new products and services towards their perspectives and needs.As more companies mine real-world data (from wearables, social media, sensors, shopping habits, etc.) to understand consumer behaviour, the role of AI becomes increasingly important in bridging the gap between insights and developing new solutions. 

It could also mean that consumers could be nudged to consume more trendy foods and beverages purchased by digitally savvy consumers, whilst healthy and affordable alternatives could be presented as less desirable. This means that any new product development would risk benefitting only a small portion of the population. We recently published a white paper on this critical topic,entitled “The ethics of AI at the intersection of nutrition and behaviour change” 

 

Widening inequality

Recent studies have highlighted significant challenges that could impact the equitable implementation of AI in health and food-related applications. A European study found that nearly 40% of the population faces challenges with digital literacy (13), while research in the UK revealed concerning levels of health inequalities stemming from lack of access to technology. Compounding these issues is the fact that many data scientists lack formal training in ethics. This is a dangerous mix of ingredients (excuse the pun) with serious consequences, particularly in the realm of food and health. It's precisely these kinds of challenges that the AI Act aims to address, seeking to ensure that AI applications in critical areas like health and nutrition are accessible, equitable, and ethically sound.

 

Spread of nutrition & health misinformation

In addition, the spread of misinformation is the most challenging threat of this generation, according to the WHO (World Health Organization). Even the nutrition and health industry is not spared, with a recent study concluding that only 2% of nutrition and health information found online is accurate (15).

The AI Act is therefore timely in its aim to prevent the potentially harmful use of AI technologies, avoiding the need for difficult and costly corrections in the future.

 

When to consider whether the AI Act applies to Food & Beverage and ingredient brands?

For companies and brands that employ AI, the AI Act applies if you:

  • Use an AI chatbot to communicate with consumers.. The majority of AI applications currently available will not be classified as high-risk, but will have to comply with transparency requirements and EU copyright law.
  • Use AI (eg. ChatGPT) to create social media content (video, text, audio). 
  • Use generative AI to create advertisements 
  • Use biometric data or health data to generate personalized advice and recommendations. This use can be considered high-risk depending on the context and application.
  • Create personalized recipes and meal plans
  • Use AI in an e-commerce tool that recommends specific products for purchase. This is particularly relevant if the recommendations are based on sensitive data or impact health significantly.

It is important to note that the EU AI Act was developed for wide application, so industry-specific guidance is still limited. This means areas that apply to Food and Beverage brands may become apparent at a later date.

 

AI risks widening inequality when AI development and datasets primarily come from digitally literate groups with greater access to technology, potentially skewing new products and services towards their perspectives and needs

 

Why should companies care?

The AI Act came into force on the 1st of August (2024) which means that companies are legally bound to comply and need to take the AI Act seriously. The use of general-purpose AI, such as ChatGPT and other Generative AI tools,will need to be compliant by next August 2025. Companies need to demonstrate how they comply or risk facing fines.

Beyond fines, companies can use this time to consider their ethical practices with regards to sustainability and social impact that go beyond the GDPR (2018) regulation.

 

What is the opportunity?

The biggest opportunity for Food & Beverage and Ingredient brands is building trust. Abiding by the new AI Act means Food and Beverage brands can develop products that not only meet consumer demands, but are transparent and offer benefits that can be enjoyed equitably to improve health. 

The second opportunity is around engagement. Consumers, especially Gen-Z and millenials, are more concerned about how their choices impact people and the planet. Abiding by the AI Act gives brands the opportunity to demonstrate how they develop products and share knowledge and information in an ethical way. Brands are more likely to get engagement when consumers trust them.

The final opportunity is around impact. Brands have the opportunity to impact the health of society by leveraging AI to nudge consumers towards sustainable choices that are nutrient dense and are culturally relevant (EAT LANCET).

 

In summary, 

The AI Act is the latest regulatory framework that will impact Food & Beverage and Ingredient brands. Companies should be clear on the risk level they fall into and be transparent about how, why and when AI is employed, even if it is used for a chatbot or social media content. The AI Act aims to ensure that products and services are developed in an ethical way in order to build trust and transparency. Food and Beverage brands who are increasingly at the intersection of food, pharma and health need to be particularly mindful that there will be increasing scrutiny in how AI is leveraged, in order to protect the public from unintended harm. Companies need to have the relevant processes and documentation in place in order to avoid fines and suffer reputational damage. There are many ways AI can employed in the Food & Beverage industry, however, executives need to be clear on how.

 

What should companies do next?

There are several actionable steps companies can take to comply with the AI Act, however expert advice should always be sought.

  • Ensure that all internal staff are familiar with the AI Act 
  • Assess where the Food & Beverage company sits in terms of the risk-level. Completing the AI Act compliance checker is an easy way to do this https://artificialintelligenceact.eu/assessment/eu-ai-act-compliance-checker/
  • Conduct an internal audit of data flows:how data is transformed and who is accountable for all data elements
  • Develop an ethical AI code of conduct as well as the relevant documentation to demonstrate compliance
  • Find experts who understand regulation, personalized and precision nutrition, and technology, to advise on the necessary steps to future proof the business. 
  • Learn more about Qina’s new services.

 

Qina has recently partnered with Lumiera, a boutique advisory consultancy in the area of Responsible AI. Together we help and support companies to understand and operationalize the principles of the AI Act.

We have developed an FAQ document for those new to the AI ACT, get in touch for more information or  book a call 

 

References

  1. Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (The EU AI Act) https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32024R1689 last accessed 27th September 2024
  2. Buolamwini, J. & Gebru, T.. (2018). Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification. Proceedings of the 1st Conference on Fairness, Accountability and Transparency Proceedings of Machine Learning Research 81:77-91 Available from https://proceedings.mlr.press/v81/buolamwini18a.html.Last accessed 27th September 2024
  3. Idowu, J.A. Debiasing Education Algorithms. Int J Artif Intell Educ (2024). https://doi.org/10.1007/s40593-023-00389-4 Last accessed 27th September 2024
  4. Chen, Z. Ethics and discrimination in artificial intelligence-enabled recruitment practices. Humanit Soc Sci Commun 10, 567 (2023). https://doi.org/10.1057/s41599-023-02079-x Last accessed 27th September 2024
  5. Advances in Facial Recognition Technology Have Outpaced Laws, Regulations; New Report Recommends Federal Government Take Action on Privacy, Equity, and Civil Liberties Concerns available at https://www.nationalacademies.org/news/2024/01/advances-in-facial-recognition-technology-have-outpaced-laws-regulations-new-report-recommends-federal-government-take-action-on-privacy-equity-and-civil-liberties-concerns
  6. Artificial Intelligence: Development, risks and regulation  available at https://lordslibrary.parliament.uk/artificial-intelligence-development-risks-and-regulation/
  7. Food waste (United Nations) https://www.un.org/en/observances/end-food-waste-day
  8. Nuritas https://www.nuritas.com/publications
  9. Eatlove https://www.eatlove.is
  10. Spoonguru: https://www.spoonguru.com
  11. Zoe – https://joinzoe.com
  12. The ethics of AI at the intersection of nutrition & behaviour change https://www.qina.tech/blog/health-digital-ethics-of-ai-nutrition-behaviour
  13. https://www.healthcareitnews.com/news/emea/40-people-europe-face-challenges-digital-literacy
  14. Joe Zhang, Jack Gallifant, Robin L Pierce, Aoife Fordham, James Teo, Leo Celi, Hutan Ashrafian - Quantifying digital health inequality across a national healthcare system: BMJ Health & Care Informatics 2023;30:e100809.
  15. https://business.dcu.ie/dcu-and-myfitnesspal-study-on-social-media-health-and-wellness-trends-highlights-urgent-need-for-digital-health-literacy/
  16. https://artificialintelligenceact.eu/