Navigating ethical challenges in an AI-enabled food industry

Abstract

AI is rapidly being adopted across various sectors, transforming roles by eliminating repetitive tasks and enabling more productive use of time, while businesses experiment with AI to gain a competitive edge. The article highlights that European legislation, such as the new AI Act, aims to ensure the responsible and ethical use of AI, mandating transparency and compliance for all businesses operating within its jurisdiction.

‘AI is coming to take your job!’ It's a phrase we’ve probably heard on countless occasions, and no doubt given some serious consideration. After all, according to BSI's research, 81% of businesses globally are investing in AI⁽¹⁾. We are seeing mass adoption, adaptation, and experimentation across almost every sector, as businesses seek to achieve efficiencies and improvements or look to gain a competitive edge.

AI has already changed some roles while making others redundant, often eliminating mundane, repetitive tasks, and enabling the time saved to be used more productively⁽²⁾. This is likely to continue as businesses experiment with and explore how AI can best benefit their processes and procedures to gain them first-mover advantage in a highly competitive market.

While some roles currently undertaken by humans are expected to disappear – according to a BSI study, 31% of business leaders across nine sectors think AI will replace specific job functions. Research suggests that by 2029, 80% of human decisions will not be replaced by AI, but will be augmented by generative AI (GenAI)⁽³⁾.

AI is quickly making significant inroads across business, and every organisation today faces the challenge of adopting and implementing AI – transparently, responsibly, and ethically. BSI research shows that globally, 83% of businesses say it's important for their organisation to inform others in the supply chain about how AI is being used in the business or their future plans to use it⁽¹⁾.

The businesses that succeed in the future will likely leverage the benefits of AI to improve many aspects of their operation – and European Union (EU) regulators have recognised this, introducing new legislation in the form of the AI Act⁽⁴⁾, which is designed to govern how businesses use AI.

This means that partner organisations, suppliers, clients, and customers are likely to scrutinise how the businesses they engage with use AI to ever greater degrees.

The pace of technological advances can be such that legislation and even guidance are playing catch-up. This is only to be expected because, until the full scope of new technology is appreciated, it can be difficult to spot potential pitfalls.

This has been true of AI. However, with the introduction of the AI Act, Europe is leading the way in legislating how it should be used.

The new AI Act from the EU is the first-ever legal framework for AI⁽⁵⁾, and has been created to regulate the development and deployment of AI, aiming to establish trustworthy AI in Europe and beyond.

‘The responsible and ethical use of trustworthy AI is going to be governed and regulated by this new legislation, which every business with any impact in Europe has to comply with, no matter where they are located,’ explains Richard Werran, Global Director of Consumer, Food and Retail, BSI and a Fellow of the Institute of Food Science and Technology.

‘Trustworthy’ is the key word here.

After all, trust is an essential commodity in business – particularly when it comes to AI⁽⁶⁾.

BSI and Censuswide research found that 77% of respondents believed trust in AI is necessary in supply chain management, while KPMG research shows that 61% of people are ambivalent or unwilling to trust AI^{(7, 8)}.

That's no surprise.

A plethora of reports have detailed worries over a number of factors related to AI, including disinformation, safety, security, bias, hallucinations, the ‘black box’ nature of the technology, as well as ethical concerns⁽⁶⁾.

Since going mainstream, we’ve seen AI used in a variety of positive and negative ways.

On the positive side, AI has been utilised for numerous beneficial purposes, such as the early detection of illnesses and diseases⁽⁹⁾, aiding scientists in better understanding and modelling climate change scenarios⁽¹⁰⁾, and protecting biodiversity by creating algorithms to analyse data from at-risk ecosystems⁽¹¹⁾.

On the negative side, however, AI has also been used to increase the sophistication of cybercrime⁽¹²⁾, manipulate society through AI social media algorithms⁽¹³⁾, and power lethal autonomous weapons⁽¹⁴⁾.

While companies across the world have experimented with AI, this has typically been on a departmental basis, meaning a lack of consistency in approach on an organisational level⁽¹⁵⁾. According to BSI's research, only 44% of businesses have an AI strategy.

‘AI has to be approached from the top level,’ says Werran. ‘It's not the IT team's responsibility – it impacts the whole business. ‘

‘AI specialists are required to implement, manage and evolve this, and it's essential businesses have the right knowledge to guide how they develop AI, identify potential uses and embed it into the business – and, of course, how it is then monitored, reported on, evaluated and evolved.’

For businesses, the opportunity at this still relatively early stage of AI implementation is to build a framework that meets current legislative requirements and can be trusted – by employees, clients, business partners and customers. This means a framework that enables businesses to confidently embrace all of the possibilities that AI presents – while complying with all ethical expectations.

AI is an enormous opportunity for the food industry. Throughout the industry, AI and predictive analytics are an increasingly important element of how we work⁽¹⁶⁾, and there are countless innovative ways in which this technology is being explored, trialled and embedded to create significant improvements. Some of these improvements have the potential to offer an opportunity to positively impact the bottom line, some are designed to drive greater customer satisfaction, and others are intended to literally save lives.

Used in these ways, AI offers the opportunity to have a positive impact on societies and communities and act as a force for good. It could help embed trust, build confidence, monitor systems and processes, and create standardisation.

For example, integrating AI and machine learning in quality control and assurance can improve efficiency and accuracy, as well as uphold the highest standards of quality and safety in food production and distribution⁽¹⁷⁾.

Equally, it can proactively alert businesses to food safety incidents with real-time monitoring of social media, enabling timely intervention⁽¹⁸⁾.

AI can also be deployed to actively monitor critical control points (CCPs) and control points (CPs) within a Hazard Analysis and Critical Control Points (HACCP) plan⁽¹⁹⁾. By analysing trends and similarly integrating live supplier data feeds, organisations can manage and appreciate a fuller picture of food and the broader safety risks, in real time⁽²⁰⁾.

AI can help consumers and restaurants better manage food consumption and requirements by tracking inventory and recommending recipes using available ingredients⁽²¹⁾. Retailers, meanwhile, have the opportunity to use AI to better understand demand, anticipate fluctuations, ensure only food that is most likely to be sold is stocked, enabling food waste in the production cycle to be monitored and reduced^{(22, 23)}.

By monitoring consumer feedback on social media, innovative new products could be rapidly brought to market. Further, AI could also help food businesses create products with health benefits and healthier alternatives, as well as streamline a whole host of other tasks, including admin, sales and marketing. For example, in some parts of business AI is increasingly being used to filter job applicants, determine the best-fit candidates and reduce the time taken in the recruitment process⁽²⁴⁾.

In a retail environment, AI is being used to help customers navigate towards particular products (and even attracting consumers into stores for particular products in the first place) via personalised allergy alerts, recognising fruit and vegetables on scales, smart checkouts and recipe information⁽²⁵⁾.

The case for AI to improve and enhance the food sector is clear – however, it's essential that AI is neutral in order that it ultimately becomes a force for good. Therefore, the use of AI comes with substantive potential ethical considerations and risks, which must be managed thoughtfully, deliberately and above all, strategically.

While AI today is said to have an equivalent IQ of 155⁽²⁶⁾, (Einstein's was 160), it's certainly not infallible and nor are the humans using it. There have been numerous examples of it constructing its own information and presenting it as convincing fact – for example, giving customers incorrect information and creating fictional court cases⁽²⁷⁾.

AI's reasoning and logic can be flawed, and research from Stanford University in May 2024 documents the lack of transparency⁽²⁸⁾, which makes it challenging to understand how it reaches certain conclusions or makes particular decisions. It also lacks the empathy and creativity of humans, meaning it's important to approach any major AI project with this context in mind.

The nature and amount of data in the food industry is vast, as are the ethical considerations.

From the treatment of workers and socioeconomic issues to geopolitical challenges, supply chain fluctuations and public health, the food industry has an abundance of ethical touchpoints to navigate and manage.

When implementing AI, into which vast quantities of data are fed, thinking about the ethical implications of both the implementation and the impact it may have on all parties involved can help ensure a trusted system.

‘A business's reputation is vitally important, and the serious risk with AI is that it can be used unethically,’ says Werran.

‘AI's output is based on its own interpretation of the data and information that's inputted – the algorithms have been built by humans, so there are numerous opportunities for biased and unethical interpretations to emerge if this isn’t managed correctly.’

The ethical implications of using AI in the food industry include:

• Cyber security risks

AI systems embedded in food and agricultural production could be hacked.

• Environmental impact

AI may not understand the context – for example, the long-term consequences of damage to sensitive ecosystems in favour of maximising crop yields.

• Data privacy

Who owns the data used by AI and how will it be shared or privacy protected?

• Equitable access

Use of AI can vary between businesses and geographies, creating potential imbalances.

• Bias and fairness

Data and information fed into AI can have inherent bias, therefore the output will also be biased.

When embedding AI into your organisation, it's essential that all ethical concerns are managed and addressed.

What would the consequence be, for example, if it transpired your AI recruitment processes were treating a group of people from one ethnic origin differently to another? Or prices were being inflated during times of day that people from lower socioeconomic backgrounds were shopping?

Even in its relative infancy, AI has already been connected with a number of high-profile cases along the lines of the above, creating major challenges for a number of businesses. Brands can build trust in the long term by proactively considering all the risks and putting the right guardrails in place from the outset.

While an individual may not directly be responsible, the negativity and loss of brand trust scenarios like these generate would be significant.

As Werran says, bias is not the only thing to be aware of. ’The ethical considerations around AI are much wider.’

He adds: ‘Solving that bias is a fundamental challenge, given its impact and the ability of AI to scale it. However, issues including fairness of outcomes, transparency, explainability, protecting privacy, maintaining user autonomy and improving prediction accuracy all need to be considered.’

AI presents a huge number of opportunities for businesses, and to take full advantage while navigating the ethical challenges, it's essential that the right guardrails and frameworks are in place.

The EU's AI Act is the first significant regulation on how businesses can use AI, and it applies to organisations across Europe. Those who perform any business in Europe will need to ensure they can comply with the forthcoming legislation.

Being able to demonstrate how AI is being used within a business will become a prerequisite for organisations wanting to work with other companies.

Those businesses that can demonstrate that AI is being deployed objectively, safely and ethically within their organisation are likely to build trust and therefore be more attractive business partners. Certification verifying this could provide that reassurance and could ultimately become a contractual obligation. Either way, it could provide an important way for businesses to protect their brand and demonstrate they are using AI in a safe and ethical manner.

The world's first AI management system standard (ISO:IEC 42001:2023) specifies requirements and provides guidance for establishing, implementing, maintaining and continually improving an AI management system for organisations of any size.

It establishes guardrails within which the organisation can work, as well as being designed to help businesses acquire the knowledge, skills and behaviours to effectively explore, embed and use AI for long-term success.

By maintaining and demonstrating compliance and promoting accountability, possessing certification can enable businesses to formally demonstrate compliance with both legal and ethical obligations, helping to meet customer, staff and stakeholder expectations.

‘The good news is an AI management system standard provides an auditable framework for organisations to build AI into their business as an extension of their existing management system,’ explains Werran.

‘By way of its high-level structure, it is designed to dovetail with other international standards, and over time, I believe that stakeholder power will demand businesses demonstrate compliance through certification.’

The opportunities for AI to impact every aspect of the food system – from optimising crop yield and increasing production efficiency to reducing food loss and waste and world hunger – are immense. That's why the correct implementation, development and use of AI is going to be essential.