Publications

In order to bring safe and reliable applications into use, the European Commission, in its proposal on the regulation of AI systems in April 2021, advocated regulating AI systems according to their risk potential and classifying them into four risk levels, from minimal risk (no need for regulation) to unacceptable risk (prohibition of use). In the white paper „Criticality of AI systems in their respective application contexts – A necessary but not sufficient building block for trustworthiness“, experts from the working groups IT Security, Privacy, Legal and Ethical Framework as well as Technological Enablers and Data Science deal with selected contents of this EU proposal: namely, which criteria can be used to determine in which cases the use of AI systems should be regulated from the outset and when this is not necessary.

This paper presents complementary findings from a micro-study. The objective of the microstudy was to gain data from business practice in order to complement the elaborations of the white paper with empirical findings. By means of an online survey that addressed company specialists and managers (N = 50), requirements for AI-specific competencies and company offerings for AI qualification were elicited, among other AI related topics. The survey was based on an extensive literature research, from which distinguished application scenarios for AI-supported office and production work were derived; they formed a subject- specific starting point for the survey. The results of the online survey were commented on by a panel of experts during a public webinar of the Plattform Lernende Systeme in Fall 2021.

The use of AI systems in companies, for example, puts collaboration between humans and technology on a qualitatively new level. Certain physical and operational activities as well as cognitive tasks will decrease in importance, while unpredictable tasks requiring communication skills, creativity, or reflection as well as a high degree of flexibility will increase in importance. The use of self-learning systems will therefore also change the competency requirements for employees. New competencies will be required in all industries and domains for the development of AI systems and for dealing with AI technologies for employees with different roles. These changed competence requirements affect technical and social dimensions in equal measure.

Fake news, disinformation campaigns, upload filters or even election recommendation apps are no new phenomena. However, the use of AI systems increases their efficiency and significance. In addition, the use of artificial intelligence (AI) does often not meet the transparency requirements of a democratic public and machine action increasingly replaces human action. The discussion about AI is therefore intense and occasionally takes on a „hype character“, oscillating between two extremes: far-reaching hopes for better systems on the one hand and diverse fears such as loss of control, surveillance, dependency and discrimination on the other. The debate‘s „hype character“ can be observed in the relationship between democracy and AI – especially with regard to individual opinion-forming in the context of elections.

In this white paper, experts from the working group Technological Enablers and Data Science of the Plattform Lernende Systeme outline various measures on how the transfer of AI knowledge can succeed in the interest of all stakeholders involved. This means transferring interesting research results to innovative practical AI-applications with economic success.

Future mobility will be digitally networked and provide individual, customised mobility services. Artificial intelligence (AI) can make an important contribution here, on the one hand by relieving infrastructures, the environment and resources in a sustainable and resource-efficient way, and on the other hand by guiding travellers to their destination in a time-saving and flexible manner. This can be done conveniently with the intelligent travel assistant as an app on the smartphone or laptop. It bundles and analyses offers in the background and suggests individual travel routes that cover environmentally compatible and sustainable mobility.

Whether in transportation, logistics, individual transport, or local public transport – transport is achieving ever higher levels of automation thanks to Artificial Intelligence. Automated driving can help increase traffic safety, optimize traffic flows, and reduce pollutant emissions. Increasingly powerful AI and machine learning techniques are improving automated driving technology so that it works in real-world tests in more than 99 percent of situations. A residual risk for possible misbehavior occurs in so-called edge and corner cases. AI systems may not be sufficiently trained and tested for these rarely occurring special cases. To tap the potential of cross-industry fleet learning, the experts of the working group Mobility and intelligent Transport Systems of Plattform Lernende Systeme therefore propose the establishment of a collaborative AI mobility data platform. This platform should enable the exchange of mobility data and contribute to risk minimization in automated driving.

Whether in space, in the deep sea or in disaster areas - operations in such hostile-to-life environments represent a major challenge with considerable risks for humans. Self-learning systems can help to reduce hazards and risks for humans or to make such environments accessible in the first place. Mission configurations, the degree of autonomy of self-learning systems and the intensity of interaction with humans can vary considerably. A good division of labor is critical to the success of collaboration between humans and self-learning systems. The working group Hostile-to-Life Environments of the Plattform Lernende Systeme has identified and investigated key requirements for this division of labor between humans and self-learning systems as well as for the competence of the self-learning systems in the respective situational context of application.

Whether through networked mobility spaces or the optimization of traffic flows, logistics and transport processes – Artificial Intelligence (AI) can make an important contribution to the mobility of the future, improving existing business models and enabling new ones. In recent years, decisive progress has already been made toward autonomous mobility. AI can help make transportation systems smarter and fit for the future. At the same time, AI-based mobility concepts enable completely new business models, for example for digital platforms enabling numerous companies of all sizes and industries to cooperate in the future. Experts from the working group Business Model Innovations and the working group Mobility and Intelligent Transport Systems of the Plattform Lernende Systeme have analyzed various stakeholder groups in fictitious logistics and travel scenarios to investigate new AI business models and the role of AI-based platforms for travel and transportation.

The certification of AI systems can help to strengthen trust in technology and application. And thus increase AI application. The goal of a successful certification process should basically be to guarantee standards and at the same time avoid overregulation and enable innovation. Led by two working groups IT Security, Privacy, Legal and Ethical Framework as well as Technological Enablers and Data Science, members of all working groups of the Plattform Lernende Systeme, together with other guest authors, have defined criteria that can be used as orientation for the certification of AI systems on the one hand and can support the decision on the necessity of certification on the other. The experts also provide an overview of efficient infrastructural and organisational requirements for certification. This takes up, at the same time, the current state of the discussion and connects to the already published discussion paper Certification of AI systems.

Artificial intelligence (AI) offers great potential for companies and their employees – whether through improved work processes and relief or digital business model innovations. At the same time, the change in companies must – and can – be shaped together and the challenges in the use of AI systems must be solved. This is the only way to overcome challenges and negative side effects in the use of AI systems. The overall aim is to create a new relationship between humans and machine, in which people and AI systems work together productively and the respective strengths are emphasized.

Artificial intelligence (AI) offers great potential for companies and their employees – whether through improved work processes and relief or digital business model innovations. At the same time, the change in companies must – and can – be shaped together and the challenges in the use of AI systems must be solved. This is the only way to overcome challenges and negative side effects in the use of AI systems. The overall aim is to create a new relationship between humans and machine, in which people and AI systems work together productively and the respective strengths are emphasized.

Whether satellite images as data sources for navigation systems or vacation photos on social media platforms – unimaginably large amounts of new data are generated every day. Data has therefore become a central resource in our increasingly digitalized world. Comprehensive data management and the ability to make data accessible for analysis in the first place are thus an important prerequisite for gaining valuable scientific insights and generating beneficial applications for business and society. The interdisciplinary branch of research data science – the management and analysis of data – is therefore already considered one of the most important key disciplines for science and business. The availability of data and data management is a central pre-condition especially when applying self-learning systems.

AI promises great potential for the healthcare system – for prevention, patient-specific diagnoses and therapies. Physicians and nursing staff can be effectively supported in their work in the future by AI-based assistance systems. The Plattform Lernende Systeme has already dealt with various aspects of the introduction of AI-based assistance systems in the healthcare system in several publications. All technological advances must focus on the needs of those affected. A central concern of the working group Health Care, Medical Technology, Care of Plattform Lernende Systeme is therefore to also take the perspective of patients into account when discussing the opportunities and challenges of AI in the healthcare system.

Ethical values and principles are important for many people, even in the economic context. This is especially true for the development and application of Artificial Intelligence (AI). Even if the possibilities offered using AI systems in the various fields of application are viewed positively by many, there are still some concerns regarding this technological innovation. A lack of trust is a major reason for the lack of acceptance of AI systems. The working group IT Security, Privacy, Legal and Ethical Framework of Plattform Lernende Systeme has developed a guideline on the challenges that need to be considered when developing and using AI systems responsibly. The guidelines also outline measures used to implement an ethically reflective technology development and application process.

Artificial Intelligence (AI) offers a wide range of potentials for safe, autonomous and self-determined work as well as attractive and competitive jobs. For example, AI-based assistance systems can relieve workers of strenuous or dangerous tasks and support them in complex processes and decisions. At the same time, AI systems are changing the interaction between humans and technology in the realm of work. In the future, humans and machines will interact even more strongly – and in different ways – than in the past since Machine Learning (ML) and similar technologies enable machines to perform certain tasks independently and to learn continuously in the process.

The use of Artificial Intelligence (AI) offers a wide range of potentials for safe, autonomous and self-determined work as well as attractive and competitive jobs. For example, AI-based assistance systems can relieve employees of strenuous or dangerous activities and support them in complex processes and decisions. At the same time, AI systems are changing the interaction between people and technology in our working environment. In the future, people and machine will interact even more strongly – and differently – than in the past, since Machine Learning (ML) and similar technologies enable machines to perform certain tasks independently and to learn continuously.

The certification of Artificial Intelligence (AI) is considered a possible key requirement for promoting the use of AI systems in various areas of business and life. For many AI systems, certification can contribute to exploit the societal potential in a secure and public-interest oriented manner. A successful certification of AI systems enables the fulfilment of important social and economic principles, such as legal security (e.g. liability and compensation), interoperability, IT security or data protection. In addition, it can create confidence among citizens, lead to better products and influence the national and international market dynamics.

The use of Artificial Intelligence (AI) promises great improvements in medicine. In the future, self-learning systems can lead to better treatment results in prevention, early diagnosis and patient-oriented therapy, thus improving our health care. The use of AI systems can also help doctors and medical caregivers to improve patient care and reduce the workload on medical staff. The fictitious application scenario „With Artificial Intelligence against Cancer“ developed by the working group Health Care, Medical Technology, Care of Plattform Lernende Systeme, provides concrete examples.

Artificial Intelligence (AI) is already being used far more widely today than we might initially expect, and the associated potential for discrimination is not always obvious. Although people themselves are guilty of unjustifiable discrimination at times, they perceive the decisions taken by computer programs and software solutions to often be factual, objective and neutral. However, in reality, AI-based systems sometimes make decisions that are problematic, discriminatory or that draw distinctions without good reason. Many software systems explicitly or implicitly comprise a set of social rules for controlling behaviour, whether in the form of regulations, transactions and coordination, or access and usage rights. First and foremost, they are an effective technical means of putting systems of rules into practice. Consequently, self-learning systems have the potential to not just adopt pre-existing discrimination, but even to enhance it.

Artificial Intelligence (AI) is a catalyst for the processes of digital transformation that are sweeping through society and the world of work. AI technologies not only impact on the requirements that people have to meet and the skills they need to have, but also on the types of activity they undertake, the places they work and the way their work is organised, both within individual companies and across the entire labour market. Furthermore, Artificial Intelligence is changing the relationship between people and technology and making new types of human-machine collaboration possible. Shaping these profound processes of change means identifying ways to achieve a future (working) world that utilises Artificial Intelligence, striking a balance between safe workplaces and trained staff, and creating work that is rewarding and centred on people.

We are living in the golden age of Artificial Intelligence1 (AI). Continuing progress in algorithm development, particularly in Machine Learning2 (ML) and Deep Learning (DL), combined with the availability of huge data sets and advances in rapid, parallel computing have helped deliver breakthroughs in diverse fields of use. Applications that just a few years ago seemed to be plucked from the realms of science fiction are now already – or are soon to become – part of our daily lives. Knowledge of unimaginable breadth and astonishing depth is becoming accessible at the click of a mouse, voice-controlled assistance systems are helping us in many aspects of life, image recognition systems have achieved near-human performance levels, autonomous vehicles are increasingly becoming a reality, business models are changing rapidly and personalised medicine is supporting optimum and personalised treatment.

Technologies based on Artificial Intelligence (AI) are increasingly permeating all spheres of life. The ways in which they can improve the security of IT systems, together with the security of AI systems themselves, are essential for enabling citizens, businesses, politics and public authorities to reap the benefits of advancing digitalisation.

Artificial Intelligence (AI) has recently made breakthroughs in very different areas of application. This progress has been made possible by advances in algorithms, particularly in the case of Machine Learning (ML) and Deep Learning (DL), coupled with the availability of very large volumes of data (Big Data) and advances in fast, parallel computing (Kersting, Tresp 2019). Applications that just a few years ago seemed to be far out of reach are now already – or are soon to become – part of our daily lives.

Download a short presentation of the key findings.

A safer, more flexible and more economical way to get from A to B by road, rail or water – Artificial Intelligence (AI) can play an important part in achieving this vision. Indeed, AI-based assistance systems help make transport systems more intelligent and futureproof. This is made possible by the interplay of sensors, cameras and intelligent infrastructures and platforms that capture, manage and share traffic data. Increasingly sophisticated Machine Learning (ML) processes are being used to process captured data and derive operations to be executed either by people or by the systems themselves.

Whether in prevention, early diagnosis or selecting the ideal treatment, Artificial Intelligence (AI) and Machine Learning (ML) could soon be playing a big part in ensuring that people receive better and more personalised medical care. There is a whole variety of ways that self-learning systems could be put to use in the medical practices and hospitals of the future. For instance, doctors could use AI systems across the board to help them evaluate imaging procedures, thereby obtaining more accurate diagnoses. By using networked data, self-learning systems could derive proposals for suitable preventive approaches or treatments, thus helping medics and patients make important decisions.

Whether exploring terrain that is difficult to access or carrying out rescue operations, the use of Artificial Intelligence (AI) promises to provide helpful solutions in hostile-to-life environments. Mobile robots and assistance systems that adapt to changed situations without needing to be re-programmed can effectively support human beings with activities in hazardous environments. These activities could include fires, disaster control and deep-sea maintenance work, for example. Self-learning systems like these reduce risks for the personnel involved, speed up responses in time-critical situations and plug capability shortfalls in situations where humans are not yet able to respond adequately. In doing so, they deliver valuable social benefits. At the same time,
AI-based systems offer huge potential for research and the economy. Thanks to their ability to learn, they make operations in hazardous or poorly accessible environments much more cost-effective than manned missions – or even possible in the first place. There are, however, still a number of technical challenges that need to be met when it comes to deploying self-learning systems in hostile-to-life environments. These include ensuring long-term autonomy and autonomous learning in unknown environments. Another important area involves shaping the interaction and cooperation between autonomous robots or assistance systems and people.

Intelligent virtual assistants that can answer straightforward questions or carry out simple instructions are already to be found in smartphones, tablets and millions of households. As yet, however, these systems are not capable of mastering the complex daily work involved in journalism, administration or consulting, for example, as they still cannot process huge volumes of information, handle different subjects, sources and means of communication or contact a variety of different people. In a few years, self-learning assistance systems that are intuitive to use will support knowledge workers in their everyday tasks.

Manufacturing companies already use a range of robotic systems – from gripping arms to lightweight robots – that automate factory processes and make the jobs of the people working on the assembly line easier. Each of these machines takes on strictly defined repetitive production steps, requiring complicated modifications whenever any changes are made, and they are therefore primarily suited for mass production from a financial perspective. They can only be used with the support of comprehensive expert knowledge. In the future, humans will direct robots in factories and, depending on current demand, teach them new skills during the production process. These robotic tools with learning capability will work safely hand-in-hand with the people on the assembly line.

There are already more than 1,300 offshore wind turbines dotted along Germany’s coastline, with more currently being built. Their foundations reach down to 50 metres below the water’s surface. This presents a challenge during inspections, servicing and repairs, making work on underwater structures elaborate, expensive and dangerous for the divers involved. Within a few years, self-learning robotic assistance systems could help humans to inspect, service and repair offshore turbines and other underwater infrastructures. These autonomous underwater vehicles (AUVs) deploy Artificial Intelligence methods.

In just a matter of years, self-learning systems will be able to provide rescue workers with effective support during operations in life threatening environments. Whether remote-controlled or autonomous, these systems can undertake a whole range of tasks – from risk prevention and defence measures to repairing damage and providing emergency aid.

Cancer is the second leading cause of death in Germany after cardiovascular diseases. A tumour in the lung is one of the most common types of cancer, with only around one in five patients surviving the first five years after diagnosis. In the future, doctors will be able to draw upon Artificial Intelligence (AI) to combat lung cancer – from the initial screening and diagnosis to treatments and follow-up care. For patients, this will open up new, personalised and highly effective treatment options that will significantly improve their prognosis.