Special Committee on Artificial Intelligence in a Digital Age

AIDA Working Paper on

AIDA
‘AI, Agriculture and Food Security’
following the AIDA/AGRI Public Hearing on 14 June 2021

June 2021 | WWW.EUROPARL.EUROPA.EU - @EP_ARTIFINTEL

Introduction

The Special Committee on Artificial Intelligence in a while the second panel focused on how AI can be
Digital Age (AIDA), in co-operation with the Commit- used in agricultural development and in achieving
tee on Agriculture and Rural Development (AGRI), food security.
organised on 14 June 2021 a public hearing on “Ar-
tificial intelligence in agriculture and food securi- In the context of the European Commission’s plans on
ty”. The hearing explored the implications of AI in the the future of agriculture in the European Union, which
agricultural sector by holding two panel discussions include the Common Agricultural Policy (CAP)2, the EU
bringing together AI experts from across government, Green Deal3, the Farm to Fork Strategy4 and the Biodi-
industry, academia and civil society. The key takeaways versity Strategy for 20305, looking at the potential role
from both panels are summarised below, followed by of emerging technologies can be important. In addi-
contributions on the outcome of the hearing made by tion, the Coordinated Plan on Artificial Intelligence6,
the European Parliament political groups. A video re- published by the European Commission in April 2021,
cording of the event is available via the AIDA Commit- covers actions to support the development of AI sys-
tee website1. tems for sustainable agriculture.

The first panel of the hearing was guided by the ques- In relation to the hearing, AIDA Chair Dragoș Tudorache
tion of how AI can promote the transition towards highlighted the following points: “Artificial intelligence
a smarter and more sustainable future agriculture, has no purpose in and of itself. Its role, as a sophisti-

2
cated technology that uses increasingly complex data ply chains, and increase the quality of the agricultural
to make increasingly accurate predictions, projections, products we consume. But this is not just about a lin-
and in some cases decisions, is to improve the way our ear increase in efficiency; by researching and deploy-
economies and societies function. Using AI in agricul- ing AI technology in agriculture and interdependent
ture is ultimately, just like using it in any other domain, domains we are also increasing our strategic resilience
a way to improve the well-being of our citizens” . and reducing our impact on the environment, making
good on Europe's ambition to become a global player
“AI in agriculture will help us make more efficient use and model for the world” he added.
of the land, optimize our resource use, shorten sup-

1 AIDA website, https://www.europarl.europa.eu/committees/en/aida

2 The future of food and farming - Communication on the Common Agricultural Policy post-2020
3 Communication and roadmap on the European Green Deal, COM(2019) 640
4 Communication - A Farm to Fork strategy for a fair, healthy and environmentally-friendly food system, COM(2020) 381
5 Communication - EU Biodiversity Strategy for 2030 Bringing nature back into our lives, COM(2020) 380
6 Communication - Coordinated Plan on Artificial Intelligence (COM(2018) 795

3
Key takeaways:
Panel 1:
The first panel focused on the question of “how can
AI promote the transition towards a smarter and
more sustainable future agriculture?”, with the par-
ticipation of four expert panellists:
• Dr Sjaak Wolfert, independent researcher associ-
ated with Wageningen University & Research, The
Netherlands
• Dr Francesca Hennig-Possenti, Chair of the arti-
ficial intelligence group at CEMA, Senior in-house
lawyer at John Deere GmbH & Co. KG Germany
• Dr Steven Davy, Head of Division, Programmable
& Autonomous Systems (PAS) at Walton Institute for
Information and Communication Systems Science that AI technologies are still under development and
(WIT), Coordinator of H2020 CYBELE that they can pose ethical challenges, however, she
encourages further development of AI applications in
• Pieluigi Londero, Head of Unit, Implementation agriculture.
Support and Integrated Administration Control Sys-
tem, European Commission AI can bring a more sustainable future for agricul-
ture, but first, the major obstacles need to be tack-
Applying AI to the domain of agriculture and food se- led in order to unlock its full potential, according to
curity requires a paradigm shift, including the adop- Steven Davy. Some of the main challenges of employ-
tion of new techniques, technologies, holistic working ing AI applications in agriculture are: lack of concrete
methods and multidisciplinary views. Sjaak Wolfert investment, reluctance to embrace new technologies
stated in his presentation, that a responsible innova- (mainly due to cost of adoption) and gaps in skills and
tion approach must include technical robustness, knowledge related to AI and the agri-food sector. Mr
human-centric design, explainable screening and Davy stated that one approach to tackling these prob-
auditing, as well as socio-economic feasibility, so- lems is to heavily invest in training within the agri-food
cial acceptance, and ethical desirability. Mr Wolfert sector, especially with the aim of raising awareness
mentioned three groups of disciplines that should about how AI can lead to sustainable and transforma-
support the use of new digital solutions: data science, tive changes in the sector. Mr Davy also underlined
AI, and the ICT sector, which are meant to develop that the uptake of AI-powered technologies should
smart algorithms and organise sound infrastructures be incentivised by the EU for agribusinesses, but
for data exchange, followed by business modelling, not without ensuring responsible, ethical and
governance, and ethics to develop solutions that are trustworthy AI. He added that some of the prerequi-
economically and legally feasible, socially acceptable sites to achieving this goal include: the need for AI to
and ethically desirable. Lastly, ecosystem develop- be explainable to farmers, more energy efficient, and
ment was also mentioned as being necessary to bring- suitable for the agriculture field.
ing the right stakeholders together and supporting
awareness-raising and communication efforts geared One of the European Commission’s political priori-
towards a wider community. If all these disciplines are ties is “Europe fit for the Digital Age”7, which aims at
harmoniously integrated, then AI solutions for agricul- strengthening Europe`s digital sovereignty and setting
ture can be developed, according to Mr Wolfert. standards in fields such as AI, cybersecurity, data strat-
egy, and digital skills, according to Pieluigi Londero. Mr
Considering that in the year 2050, we will have to pro- Londero highlighted that AI technology in agricul-
vide food for a very large number of people - up to 9 ture must be supported by data. The European Com-
billion - the agricultural sector will be facing major mission Proposal for a Data Governance Act8 is geared
challenges, stressed Francesca Hennig-Possenti. In or- towards creating a harmonised framework for data
der to adjust to the new reality, we need to grow the exchanges, making public sector data available for re-
capabilities of agriculture and engage with AI not only use, and ensuring sharing of data among businesses
in the field of robotics but also within the field of in Europe. This initiative alongside the AI legal frame-
data analysis. Ms Hennig-Possenti stated that em- work and the Coordinated Plan on AI can help farmers
ploying AI in the field of agriculture represents a ne- meet economic profitability and environmental goals,
cessity for achieving food security. She recognized according to Mr Londero.

7 Political guidelines of the Commission 2019-2024

8 Proposal for a Regulation on European data governance (Data Governance Act), COM(2020) 767

4
Panel 2:
The second panel focused on the question of “how procedure for farmers, is needed. Innovation should
can AI be used in agricultural development and in start with farmers as they are the party most aware of
achieving food security", with the participation of the dangers of AI, such as its high energy consump-
four expert panellists: tion. Considering that farmers are usually in a weak
bargaining position and that specific groups of farm-
• Juan Francisco Delgado, Executive Vice-President, ers could be further marginalised, it is better to avoid
European Foundation on Innovation (INTEC), Sena- the assumption that AI will inherently contribute to
tor for the European Parliament Union of the World the Farm to Fork strategy or the European Green Deal,
Business Angels Forum (affiliated to the G-20 for is- Ms Schimpf stated.
sues of social inclusion) and Vice President of Start-
up Europe Smart Agrifood Summit AI is not the appropriate technology to rely upon
for agricultural development or food security,
• Mute Schimpf, Food Campaigner stressed Rian Wanstreet. Due to climate change, which
has increased resource competition, the field of agri-
• Rian Wanstreet, PhD Candidate at the University of
culture is experiencing attacks on food supplies by ma-
Washington and researcher at the UW Tech Policy
licious actors. By continuing to rely on technology in
Lab, Fellow at Harvard's Berkman Klein Center and
the agriculture sector, we are at risk of undergoing
Central European University’s Center for Media and
ransomware attacks, hacktivism, alteration of data
Data Studies
with the aim of by-passing regulations or manip-
• Matija Zulj, CEO & Founder, AGRIVI ulating market prices, according to Mr Wanstreet.
Taking into account the uncertainty of whether AI will
Keeping in mind that Europe competes in a global ever be able to mitigate the carbon footprint of agri-
race to enhance agricultural capabilities, the new culture, Mr Wanstreet added that the high level of en-
AI paradigm requires determination while being ergy required by AI technologies to gather, store, and
able to protect the environment and battle global share data is detrimental to the field. Moreover, since
hunger, stated Juan Francisco Delgado. Employment real-world conditions make it impossible to train AI
is a key aspect in the field of agriculture, and intro- on consistent and clean datasets, a complete redesign
ducing a system of incentives for quality employment of agriculture to become AI friendly would be need-
should be a priority alongside mass training for farm- ed to efficiently employ applications. Mr Wanstreet
ers in AI use, Mr Delgado said. Data, a driving force for expressed doubts on whether AI in agriculture will be
AI, should be secure, transparent, and traceable and, able to empower farmers, increase food production,
according to Mr Delgado, block chain can help in this decrease input use or cut emissions, and underlined
endeavour. Considering that globally, 420,000 peo- the contradiction of the European Commission sup-
ple die every year due to food contamination, with porting the Green Deal while moving forward with
children under the age of five being particularly at adopting AI technologies for agricultural uses.
risk (according to WHO), consumers must be placed
at the heart of the food supply chain. In addition, Mr In moving towards sustainable agriculture develop-
Delgado encouraged farmers to use AI for exchanging ment and ensuring food security, AGRIVI’s Matija Zulj
information in real time, identifying the amount of fer- presented his views on the three main drivers of dig-
tilizers needed and reducing the use of pesticides. italisation in agriculture. Firstly, the business envi-
ronment of traditional open-field agriculture has the
Farmers must be included in the debate regarding responsibility of making decisions in real time with no
the development of digital farming and they must room for error. Secondly, customers are driving digi-
be aware of new technologies, highlighted Mute talisation by expressing their preferences for sustain-
Schimpf at the beginning of her presentation. Ms ability, requiring farms to apply agricultural practices
Schimpf stated that it is challenging to manage mixed that are both sustainable and which respect agricultur-
data sets and data from different sources - from farm- al quality standards. Lastly, labour shortages paired
ers, the public, and data recorded by machinery, and with the Covid-19 pandemic have drastically affected
underlined the necessity of knowing who is allowed seasonal workforce migration, meaning that farms are
to access, collect, and aggregate data, including at risk of not being able to carry out their activities. Mr
third party transfers. In this context, data transparen- Zulj stated that AI can support farmers by, for exam-
cy is crucial, and a clear process of how farmers can give ple, providing thorough analysis, bridging aggre-
and withdraw consent should be ensured, Ms Schimpf gated data or disseminating information to farms
added. Using AI is not the only way of developing quickly. He emphasized that all AI-based practices
best practices in agriculture and techniques such should be sustainable and matched with regulato-
as better crop rotation or more diverse mix of plant ry and market-driven standards.
seeds can also help, for example, in reducing pesti-
cide use. Ms Schimpf said that a bottom-up approach,
which provides data sovereignty and a democratic

5
 Disclaimer:
Contributions by Political Groups The statements herein below are drafted by the political groups of the European Parliament. The opinions
expressed are the sole responsibility of the author(s) and do not necessarily represent the official position of
the European Parliament. This document may contain links to websites that are created and maintained by
other organisations. The AIDA Committee and the AIDA Secretariat do not endorse, nor are they affiliated
with, the view(s) expressed in the said statements nor through the said websites.

Group of the European People's Party (Christian Democrats)

The agricultural sector will be facing significant challenges in the upcoming years, especially consider-
ing that by 2050 we will need to provide food to up to 9 billion people, a significantly larger number of
people compared to today. On these challenges, artificial intelligence technology could play a major
role, enabling farmers to increase food production, decrease unnecessary costs, optimise resources
and/or cut emissions.
In order to achieve the best technological advancements in the agricultural field, farmers should be
aware of the new opportunities that technologies provide. Farmers must be included in the debate
regarding the development of digital farming.
Collecting and gaining the insights of data will play a key role in this technological progress. We ac-
knowledge that collecting and processing mixed data sets and data from different sources can be
challenging, especially for smaller farmers. That is why we should ensure a clear scheme, clarifying who
is allowed to access, collect, and aggregate data, including third party transfers.

Group of the Progressive Alliance of

S&D Socialists & Democrats

in the European Parliament

Group of the Progressive Alliance of Socialists and Democrats in the European Parliament
AI can contribute to more sustainable farming practices by providing innovative solutions and control
methods, and by making it possible to work more effectively and sustainably. Equal access to new
technologies is a key precondition to a fairer, more sustainable and more transparent agriculture, as
the use of AI on capital-intensive big farms could exacerbate further structural changes in rural areas.
Thanks to AI, more skilled workers could be interested in entering into agriculture, however, low skilled
workers should be able to receive necessary AI skills and digital literacy, in order to succeed in the jobs
transformation, leaving nobody behind. As farmers increasingly use AI, the companies which have
developed and operate these technologies might get access to data about the farmers’ land, quality of
soil, water resources, type of crops and animals on their holdings or on yields (for example from grass-
land, harvest volumes, or from animals in terms of animal weight, reproduction or volumes of milk).
The farmers’ interests in terms of data ownership and data protection must be safeguarded. When it
comes to food security, AI can be used by authorities to better target inspections based on risk criteria,
thus leading to a more efficient use of public resources and to a stronger enforcement of food security
standards in the EU.

The Renew Europe Group

Artificial intelligence will play a crucial role in agriculture to support the green transition and reach
our climate objectives. While dramatically improving work conditions, yields and profitability for farm-
ers, AI based monitoring, measurement and processing of agricultural data will help, for example, to
rationalize practices and reduce reliance on material inputs, such as fertilizers or pesticides, or fos-
sil fuels, by fostering the use of lighter and smaller autonomous machines. AI based applications are
already numerous and many promising precise agriculture technologies will emerge in the coming
years. However, to bring digital transition into the fields, we need to guarantee a fair access to these
new technologies and promote them. As 40% of rural areas in the EU have no access to high-speed in-
ternet, there is a dire need to support a faster deployment of broadband infrastructures. We also need
to provide farmers with the necessary support and incentives for the uptake of these game changing
technologies, for instance by ensuring genuine rewarding of collection and sharing of data.

6
Group of the Greens/European Free Alliance
Before applying AI to the agricultural sector, we need to assert the kind of agriculture we want. The
current model is not sustainable for the environment nor farmers. Giant monocultures rely on non-re-
productive seeds, pesticide and fertilizers and deplete the quality of the soil, biodiversity and food re-
silience. Farmers depend on crop giants upstream to get new seeds, chemicals and indebt themselves
for costly equipment and on multinational retailers downstream who control the price of crops. Add-
ing AI to such a system will only worsen the degradation of biodiversity, the situation of dependency
of farmers by handing data power to third parties and add the high-energy consumption and rare
metal depletion to the list of adverse environmental impacts. Instead, AI could model agricultural re-
silience, crop diversity and increase food quality and supply in permaculture. With 1/3 of food wasted
every year, improving the management of the food chain would also improve food security. Instead of
planning AI-piloted drones to pollinize flowers and crops in the future, we can put AI to better use than
expanding the current practices responsible for the death of bees.

The Identity and Democracy Group

Artificial intelligence in agriculture could participate in the challenge of feeding a growing population,
by improving the ratio between agricultural output (usual food) and agricultural input (land, energy,
water, fertilisers, pesticides, etc.). Through precision agriculture, the data collection on soil, weather
and crops, and their sharing with all the actors in the supply chain, we can optimise the input/output
ratio, reducing food and resource waste. Any use of AI in agriculture should be aimed at food security,
safety, and the environmental and economic sustainability of farming. AI can also be a solution for
agriculture in the developing world, especially in Africa. Basic hydroponic farming and AI could allow
developing countries to reach economic sustainability and food self-sufficiency in an environmentally
friendly way. The European Union should also be an important actor in the process, by supporting the
introduction of AI and precision agriculture in less developed countries.

The Left group in the European Parliament - GUE/NGL

The current intensive agricultural model is unsustainable for our planet. AI applications in the sector
are mirroring this dangerous path, exacerbating monocultures, especially the ones linked to high wa-
ter demand.
In a system where brokers collect agricultural data to influence the markets, the sovereignty over this
data is key. Small farmers and short-circuits risks being marginalised by this digitalisation. Automation
of agriculture risks destroying numerous jobs.
AI applications in agriculture are often praised for their potential reduction of pesticides or their ef-
ficient water management. Although it will not be sufficient to reach ecological sustainability. AI in
agriculture has yet to prove its beneficial social and environmental impact. AI is no magical solution.
Our planet and food sovereignty requires a massive U-turn in our agricultural model, with agroecology
policies, in defense of small, medium and familial agriculture models, supported by massive employ-
ment policies and targeted supportive AI applications.

7
ABOUT THE EDITOR: FOLLOW US:
Special Committee on Artificial Intelligence in a Digital Age
European Parliament WWW www.europarl.europa.eu
Directorate General for Internal Policies of the Union
BRU - KOHL Building @EP_ArtifIntel
[email protected]