NCMU's news
Round table "Best practices of applying artificial intelligence to solve social problems"
As part of the educational module "Planning Changes for Implementation of AI Solutions" of the program "Artificial Intelligence in Public Administration" a round table discussion was held dedicated to the best practices of applying artificial intelligence to solve social problems.

In the discussion, officials analyzed case studies on the use of artificial intelligence presented by scientists by leading universities (such as MIT, Harvard, ETH, TUM, and others).
  • Open Algorithms for Identity Federation (T. Hardjono, A. Pentland, MIT) The problem of data sharing (identity problem) arises from the implementation of the fixed attribute approach adopted by the consumer identity management industry. This approach provides limited information about an individual and therefore has limited value for service providers and other participants in the identity ecosystem. The Open Algorithm Paradigm (OPAL) is proposed, according to which, instead of sharing data, verified algorithms should be shared. Collective sharing of algorithms should be done through a specialized trust network. In doing so, algorithms for specific datasets should be validated against the rules of 1) confidentiality, 2) fairness, and 3) no bias.
  • AI-enabled Recommender System (I. Dolgopolova, TUM) "Without ethics, AI cannot fly" With AI, both technical and ethical issues are interconnected, and the only way to solve these problems is through interdisciplinary work that combines efforts with all relevant stakeholders, including industry, civil society, government, and academia. Working on this problem at the local, national, or even regional level is no longer enough. AI is a global problem, as is the ethics of AI. Ethically, as well as economically and socially, the ethics of AI matters. The topics of the projects under consideration -- AI ethics in 1) business, 2) autonomous robots, 3) medicine and health care, 4) online, 5) governance and regulation, 6) sustainability
  • A Layered Model for AI Governance (U. Gasser, V. Almeida, Harvard) Discusses various issues related to the governance of AI systems and presents a conceptual framework for thinking about AI governance, autonomous systems, and algorithmic decision-making processes. Researchers have reflected the complex nature of AI governance using an analytical model with three layers: the social and legal (combined into one) layer, the ethical layer, the technical layer (the technical foundations that support the ethical and social layers). These levels contain tools to address the three main problems associated with the "regulation" (in the broad sense) of AI-based technologies: information asymmetry, finding normative consensus, and governance inconsistencies. The tools contained in each of the three levels can be developed in one of three time periods: short-, medium-, and long-term perspective. On the technical level, tools can be developed in the near future, this includes the development of standards and principles for artificial intelligence algorithms. At the ethical social and legal level, in the medium to long term, states can work on specific legislation governing mature AI applications.
  • Collaborative localization of aerial and ground robots through elevation maps (R.Kasllin, P.Fankhauser, ETH) To organize situational awareness and interaction with the environment, as well as the interaction between ground and airborne robots, a localization method that allows a ground robot to determine and track its position on a map obtained by a flying robot is considered. To support invariance with respect to different sensors and viewpoints, the method uses elevation maps constructed independently from each robot's onboard sensors (the method is not applicable in planar environments). Elevation maps are used for global localization: in particular, it is possible to find the relative position and orientation of a ground robot using an aerial photo as a reference point. A particle filter is implemented to compare elevation maps, allowing multiple location hypotheses to be tested and using the robot's motion to obtain a unique solution. This allows the ground robot to use extended map coverage from the flying robot.
The round table was organized by International Innovative Institute for Artificial Intelligence, Cybersecurity, and Communications. A.S. Popov SPbGETU "LETI with support by Pavlov center "Integrative physiology for medicine, high-tech healthcare, and stress-resilience technologies" (NCMU)

The educational program "Artificial Intelligence in Public Administration" is realized on behalf of St Petersburg Corporate University in partnership with several organizations among which are ETU "LETI" and Center for Strategic Research North-West and is aimed at upskilling of public officers in the AI and related tech topics. The central issue of the module course is how to design, plan and procure AI-based applications for public and governmental needs.

Related links