Artificial intelligence in agriculture: White Paper on AI and "Special Session AGRAR: Artificial Intelligence"
AI in the agricultural sector is becoming more tangible: the European Commission published a White Paper on Artificial Intelligence on 19 February 2020. The joint project DAKIS was presented by Prof. Dr. Bellingrath-Kimura (ZALF), at the "Special Session AGRAR: Artificial Intelligence" to more than 100 experts. Prof. Dr. Reiner Brunsch, spokesman for the research association, gives statements on both events.
20.02.2020 · FV Lebensmittel & Ernährung
- Organizer of the agricultural dialogue format "Special session AGRAR: Artificial Intelligence" for politics, business, science, media and NGOs: Genius GmbH
- Lectures on 13 February 2020: among others Artur Auernhammer, Prof. Dr. Ines Härtel, Dr. Ansgar Bernardi, Jörg Migende, Prof. Dr. Engel Hessel, Prof. Dr. Sonoko Dorothea Bellingrath-Kimura
- Summary: Dr. Doreen Burdack
- Statements: Prof. Dr. Reiner Brunsch, Spokesperson Leibniz Research Association "Sustainable Food Production and Healthy Nutrition" and Head of Department Digitization & Agriculture
On 13 February 2020, more than 100 representatives from academia, politics and business met in Berlin to discuss the importance of artificial intelligence (AI) for agriculture.
The focus lied on the central question: who owns the data? For the invited speaker Artur Auernhammer, MdB, agricultural policy spokesman of the CSU parliamentary group, "digital crops" should belong to the farmer. He expressed concerns about foreign agricultural structures, and in particular he sees problems for small farmers. As a solution, he proposes the model of "machine sharing", in which technology is bought from a merger of small farmers and then shared across businesses. This would make the use of AI possible even in small-scale agricultural structures without being technically retrograde.
Statement By Prof. Dr. Reiner Brunsch: "Unfortunately, in the past year (since the hearing on digitalization in agriculture in the German Bundestag on 11.02.2019) we have not made much progress on these issues, although there is a consensus that rapid action is needed."
Prof. Dr. Ines Härtel (Europe-University Viadrina, Frankfurt-O.) sees AI as an assistance system that supports people and emphasizes that one has to face the problems of data protection (according to GDPR) with technical innovation and discusses the solution of anonymization of data. AI systems can de-anonymize them by linking data, which poses a risk and could discourage the target group from using the technology. However, purely automated decision-making processes are hardly used today, says Prof. Dr. Härtel. As a solution, she proposes to create transparency about what happens to the data, even if it is anonymized. She also calls for a new risk liability law in relation to AI solutions with an associated insurance system, as the previous liability law is very error prone. There is a multicausality, as different actors are involved in the development and use within a life cycle of an AI solution. So who is liable if damage occurs? The AI developer, the manufacturer of hardware, the distributor, the importer or the user? The current liability law does not seem to be able to answer this.
Statement Prof. Dr. Reiner Brunsch: "Europe wants to go its own way, that of trusted AI. The development of these technologies is also intended to follow the European code of values by placing the benefits for people centrally. The unlimited market mechanisms, as well as anti-democratic applications, are to be given a socially acceptable framework. It is encouraging that the new Commission is committed to this path and that AI applications are not detached from the issues of data use."
In addition, three projects were presented at the special meeting AGRAR: "House of Crops" and "prospective. Harvest" reported Prof. Dr. SonokoDorothea Bellingrath-Kimura of the Leibniz Center for Agricultural Landscape Research (ZALF) on the joint project DAKIS (Digital Agricultural Knowledge and Information System), in which the ZALF also has two other member institutes of the Leibniz Research Association, namely the Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB) and the Leibniz Institute for Innovative Microelectronics (IHP) (further information in the Annex).
Prof. Dr. Engel Hessel, Digitization Officer in the Federal Ministry of Food and Agriculture, Stressed, that AI is not new, and instead was only called differently in the past. Algorithms and decision models as well as Deeplearning solutions already existed in the 1950s. It does not assume that we will have 100% functioning systems by tomorrow. However, in order to achieve this in the future, 14 experimental fields, funded by the BMEL (Federal Ministry of Food and Agriculture), have been initiated and promoted nationwide. These are increasingly focused on technology transfer through the participation of business enterprises. The Federal Ministry of Food and Agriculture will shortly publish a new call for proposals on AI.
Statement Prof. Dr. Reiner Brunsch: "To conduct science in an interdisciplinary manner and to derive it from the societal challenges is the general goal of the Leibniz Research Association "Sustainable Food Production and Healthy Nutrition", which has been actively pooling scientific competences, developing innovative research approaches and providing a platform for dialogue and knowledge transfer between science, business and society, since 2012."
In addition to the promotion of innovation by science, it should be the state's task to create framework conditions for the use of AI and to initiate measures that aim at a digitally responsible citizen (i.e. also farmers), according to Dr. Ansgar Bernardi (DFKI).
The reason why information technologists find the development of digital solutions so "sexy" is that agriculture is not in a closed system (as opposed to production in an industrial hall, where constant and human-controllable conditions prevail). Variable weather, different soils, different genetics and behaviours of animals and plants determine agriculture and thus generate different data, linked to a certain "unpredictability". This seems to be an exciting challenge. Due to this variability and volatility, the quality of data and the accuracy of IoT in data collection (e.g. imaging evaluation, sensors, text analysis, tests, etc.) are extremely relevant. Here it is also important what kind of data it is: personal data, machine data, etc. and how it is made available (keyword: open data).
Statement Prof. Dr. Reiner Brunsch: "As an agricultural scientist, I am pleased about this development, to which we have been committed, for example, by setting up a working group "Data Science in Agriculture" at the ATB. We see the interest of IT specialists as a great opportunity to jointly address the challenges facing the agri-food industry and to develop solutions. In this way, digitalization becomes a means to an end. "
The AGRAR special session also addressed the capacity limits of data storage. As a practical solution, a purpose-oriented data compression has been proposed, in which the processing of the data is sought as early as possible (keyword: edge computing). If data is already processed in the machine and only the results need to be stored or forwarded, this significantly reduces the data volume. This approach is not practicable for science, because it is not always clear in advance where it’s going, and research results must be reproducible and comprehensible. Therefore, original data is important in research and the data storage problem is still present.
The European Commission's White Paper published on AI, provides a framework of values and legislation for the development and use of AI in Europe. Together with the European Commission's report on the impact of artificial intelligence, the Internet of Things and robotics on safety and liability, also published on February 19th 2020, important political steps have been taken.
Statement by Prof. Dr. Reiner Brunsch: "The policy framework set by the European Commission with the White Paper on AI is also very important for science. Among other things, the clear orientation towards the creation of excellent environmental conditions and trust for AI developments and applications is named as the goal. It is now up to the experts in the various sectors of the economy to identify the specific requirements in order to move closer to the greater goal of using AI to increase the common good. With the diverse competence of the partners, the Leibniz Research Association will make its contribution to helping to shape AI applications along the food value chain from agricultural production to consumers."
Contact & Further Information
Link to the European Commission's AI White Paper (Publication: 19.02.2020)
Summary Special session AGRAR: Dr. Doreen Burdack, Research Associate "Digitalization & Nutrition" at the Leibniz Research Association "Sustainable Food Production and Healthy Nutrition"