Conclusions
Together, big data and artificial intelligence will have a profound impact, providing many potential benefits across all sectors of society. NRP 75 explored a variety of ways to speed up development of new technologies and applications, and address the related societal issues. Staying in control of this evolution represents a major challenge to our public and private institutions. It requires focused efforts in many directions, including education, regulation, sector-wide initiatives and public discussions. In the following, the conclusions of the NRP 75 Steering Committee are presented, based on the knowledge gained from five years of research and on their collective insights.
The conclusions are intended to promote appropriate developments and to support measures already underway, but it is up to stakeholders to decide whether to transform them into concrete actions. They articulate needs and prospects identified from the perspective of research in data analytics and other big data related research fields. This condensate has been elaborated by the members of the Steering Committee based on the outcomes of the research projects and on their own expert knowledge and experience.
Research can produce answers to individual questions and develop specific solutions. However, this might also lead to overlaps or incompatibilities between the different approaches. It is not up to researchers to determine the societal priorities and balance; it is a matter for politicians and voters.
This Programme Résumé with its conclusions constitute a contribution by the scientific community to the formation of opinions, to political and professional debates, and to the planning of strategies and measures to develop big data applications and regulations. It is in particular addressed to actors who play a major role in defining the Swiss data space and who are therefore in a position to shape it.
Foster an appropriate environment for big data development
(1) Enhance education of big data professionals
The competent use of big data technologies calls for new knowledge and skills. Today’s IT professionals, even those educated a decade ago, may struggle with some aspects of big data such as responsible use of data, data integration and engineering, analytics, machine learning, and visualisation. There is a shortage of qualified professionals all along the value chain and in the respective fields. There is fierce competition for the best talent among academia and large IT companies and start ups. To achieve the benefits of big data in businesses and industry, in society, and in research, it is recommended that school- and university-level education in big data be expanded, such as in the form of enhanced programmes and continued education.
Universities, including universities of applied sciences, should enrol more IT students and offer tailored programmes in big data at the bachelor, master, and doctoral levels. Courses on pertinent aspects of big data should be made available in disciplines that make frequent use of big data. This may increase gender diversity in IT since big data is broader than core computer science, requiring interdisciplinary skills that also encompass societal, commercial, and legal aspects.
Continued educational offerings on big data should be expanded for IT professionals and apprentices, covering all aspects of the big data pipeline. Expertise in data collection and preparation is needed, including skills in integrating, cleaning, and linking data. Further, there is a need for skilled use of infrastructures for storing, managing, and extracting value from data, which calls for training on the rapidly evolving open-source libraries available for these purposes. It is important to master tools that visualise the outcomes of data analyses. And finally, professionals must understand the legal, regulatory, and ethical issues surrounding big data. This can be achieved by offering interdisciplinary programmes.
Since everyone in society is affected by the rapid developments in big data, education in big data must also be intensified and continuously modernised in schools, high schools, and professional apprenticeships. It would cover data science and its various facets, including law, ethics, and societal issues.
(2) Support legal and ethical advice for big data research and development projects
Many research and development projects need legal and ethical advice on aspects such as which data can be used and shared, and through which processes. Also, in case a public debate occurs in relation to a project and its findings, the project managers may be challenged to successfully communicate and convincingly demonstrate that its efforts are lawful and ethically sound. This can and should be enabled by making competent consulting and trusted audits available at affordable cost.
This type of problem is particularly acute at universities, public research institutions, and applied science institutes, as research opportunities are lost, and education programmes become less attractive. The situation is further exacerbated by the risk of negative media coverage and limited protection of university employees. Project managers need a service, either in-house or within the public administration, that provides them with competent, trustworthy, dependable, and quotable advice on legal and ethical issues as well as on communication.
How should activities be designed to be legally compliant and ethically sound? What are the reasons for possible limitations? And how can activities be communicated transparently, particularly in case of contentious public debate? A service could provide advice on such questions, easy-to-follow guidelines for practical implementation, and easy-to-understand explanations of legal and ethical considerations. In addition, it should offer audits of data usage and sharing that address the specific challenges of a project, carried out by qualified, possibly specifically certified, auditors.
(3) Enable certification of big data application properties
Big data applications have the potential to improve a wide range of processes, both in public administration and in the private sector. In some cases, applications face concerns related to fairness and biases, discrimination, ethical standards, privacy, etc. To enable the adoption of such applications, it is recommended to provide means of certifying relevant properties of such applications. This includes both the specifying pertinent properties and offering procedures that allow certification of compliance with these properties.
Numerous applications of big data are already found in non-sensitive domains, often resulting in improved performance and efficiency and in cost savings. However, they also hold potential to enable considerable benefits in sensitive settings, for example in law enforcement or social and health services. Processes for regulatory compliance and certification of compliance already exist in important areas, such as energy, construction, or trade. It is recommended to extend them to also encompass properties specific to big data applications.
Integrate big data in public and private organisations
(4) Increase the exploitation of big data technologies in the health sector
Health is a prime example of a sector where the potential of big data analytics is recognised widely by stakeholders but remains far from being fulfilled. A stronger focus on data-based management and decision making will transform current practices in the health sector, potentially improving transparency, quality, safety, efficiency, and coordination of healthcare and enhancing healthrelated competences of patients. This potential must not be left unused. The legal and ethical aspects need to be addressed for big data analytics to be used more widely in the health sector.
Switzerland has established electronic patient dossiers (EPDs) and quality measures in the health sector that demonstrate the possibility for federal initiatives in this domain. In its current form, the EPD amounts to an unstructured collection of often scanned documents that lacks structured summaries and indices, harmonisation of data formats, semantic interoperability and standardised terminology. Furthermore, it lacks compatibility with automated data analysis requirements and standardisation guidelines. Still, the EPD has enormous potential if an anonymisation layer is introduced to enable collective patient data analyses. There are privacy-preserving technologies that can help in this regard, such as data enclaves, where data is processed without being directly accessible from the outside, or federated analytics that enable local processing of distributed data to avoid compromising security. However, such technologies need further development and require committees with the relevant ethical, legal, scientific, and statistical knowledge to supervise use of the data.
(5) Strengthen policy making and evaluation with big data
The collection of data and the increased availability of advanced data analytics combine to offer a powerful foundation for strengthening the factual foundations for policy making. This makes it possible to increasingly quantify social and economic problems and to evaluate the effectiveness of policies and regulations. This potential should be realised in a way that is both responsible and beneficial.
In return, data used for the design and evaluation of policies should be publicly available, which again calls for privacy and anonymisation considerations. Altogether, the relevant offices and entities must be strengthened and expanded to handle the increasing workload and necessary communication. Procedures and mechanisms for implementing this cooperation need to be established and developed.
(6) Promote shared data collection, application benchmarks and open-source software
The availability of a variety of freely accessible infrastructure will accelerate value creation from data. To enable more open data, adoption of refined data publication policies is recommended. Likewise, better support for the creation of benchmarks and use cases for applications in different domain sciences- is warranted. Open-source software represents an attractive alternative to commercial software with expensive licences. To enable additional open-source functionality and capabilities, including, e.g., next-generation computing infrastructure and machine learning toolkits, additional funding for the development of open-source software is recommended.
Applications development can also be accelerated by having benchmarks that encompass anonymised datasets and use cases representing common scenarios in the targeted application domains. Such benchmarks can serve as references for application development and testing and for improving the accuracy and predictive capabilities of algorithms. They can enable more effective large-scale deployment and validation of big data applications.
The increased availability of open-source software can accelerate value creation from data. Thus, the creation of incentives for the sharing of tools as open-source software is recommended. For example, impact on society through the development of open-source software with wide-scale adoption should be rated on par with citation counts in the evaluation of scientific careers. This would not only promote an important public service, but might also attract international talent to Switzerland and become an overall important element in the digitalisation of Switzerland.
Update and create adequate regulation
(7) Pursue more proactive regulation of big data
While big data technologies are being deployed at a rapid pace, regulation is in its infancy and lags well behind the technological development. The lack of regulation can have adverse effects, including on democracy, on the mental health of youth, on competition as well as on innovation, for instance because of unfair advantages and reduced competition. As regulation plays a key role in avoiding such adverse effects and holds the potential to enable improved big data value creation, it is recommended that across-the-board efforts be made to accelerate the regulatory processes.
The big data divide – the asymmetric relationship between those who collect, store, and analyse big data and those subjected to data collection – is an inevitable consequence of a society that values freedom and diversity. Instead of trying to eliminate it, it is recommended that legislation identifies realistic harms that could result from the big data divide and develops legal safeguards for those who are disadvantaged.
Successful big data applications call for trust and acceptance. When putting in place frameworks under which data can be collected, analysed and used, one should not only insist on the creation of (self-regulatory) standards that balance the interests of companies and customers, but should also empower customers to make informed decisions.
Overall, it is of high importance to develop legal safeguards to compensate harm caused by the big data divide by setting standards for data collection, sharing, and analysis, enabling the protection of groups rendered vulnerable due to the deployment of big data technologies.
(8) Advance data privacy and digital sovereignty in big data applications
Deploying big data-based applications incurs risks for the privacy and related rights of individuals. Even if basic legal frameworks are available in Switzerland (new Data Protection Act) and in the EU (General Data Protection Regulation GDPR), compliance with the applicable rules is often challenging. It is recommended to raise the awareness of privacy issues and data protection rules among data scientists and engineers, data owners, and data protection officers, to elaborate comprehensive data privacy standards and to pay increased attention to the security of digital infrastructures.
A national data protection and data-related rights agenda includes numerous actors and topics. Therefore, it is important to encourage the creation of strong ties among the many stakeholders from the wide range of disciplines involved in the creation of big data applications. In addition, a methodology should be developed for establishing best practices for gathering and anonymising data, providing secure storage data, and ensuring privacy-preserving value extraction. For example, the development of data enclaves for highly sensitive data appears to be a valuable option to ensure privacy. Furthermore, the existing concept of informed consent should be complemented by specific protection mechanisms.
Various privacy-preserving techniques are being researched, but their real-world deployment will take time. Some of them could contribute to a national data privacy and data-related rights strategy, such as for example the appointment of data trustees protecting personal data of individuals, or the implementation of fairness criteria related to big data analytics to avoid discrimination. To facilitate the daily handling of privacy-related issues, further data protection guidelines should be developed. An expert competence centre such as, or within, the National Cyber Security Centre, offering a public service for addressing legal questions around privacy issues in big data deployment could be also established.
(9) Increase transnational harmonisation of regulations
Data often flows across borders, and data access from abroad and international deployment of big data-based services are prevalent. Therefore, a purely national perspective on the application and regulation of big data is insufficient. Rather, it is necessary to observe and engage internationally. Due to the numerous international organisations with their headquarters in Switzerland, Switzerland is in a unique position to support harmonisation activities of transnationally oriented institutions. Switzerland has the opportunity to demonstrate its commitment and expertise in international organisations as well as in national legislation.
While Switzerland engages actively in negotiations, further support is recommended. Similarly, it is highly recommended that Switzerland provides its input in the ongoing development of the OECD Guidelines on Responsible Business Conduct (RBC). Finally, Switzerland played an important role as promoter of the UN Internet Governance Forum in Geneva; in view of the increased tensions in the digital world, it is recommended that Switzerland makes efforts to help avoid fragmentation in the regulation of the data-driven economy.