International Partnership for Young Scientists Working on the Humanization of Artificial Intelligence

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Isaac Asimov

Modern artificial intelligence (AI) is an everyday tool in information technology widely used by people in spheres ranging from entertainment (games and creative work) to complex control and decision-making systems in business, industry, medicine and others. In most cases, artificial intelligence is created by mathematicians and programmers according to the requirements of the сustomer.

The objective of applied systems analysis for any complex task is to attempt to formalize the true aims based on the opinion of the majority (ideally all) participants involved in the problem situation and to model a system using artificial intelligence with all its many diverse internal and external connections in order to evaluate the effects of using AI as adequately as possible.

---

The Problem of Humanizing Modern AI

Ever since its beginnings, AI has been developing in two major directions: bionic (reproducing the physiological features of humans) and pragmatic (“black box” AI). The latter is linked to the creation of instruments for solving complex tasks earlier considered solvable only by humans. Moreover, these instruments can be incredibly distant from models of human thinking, and in general, do not in any way model features of the human decision-making process. It is “black box” AI that is developing most rapidly at present, and the best results and greatest prospects for the future are in this direction.

Unfortunately, such “computational” intelligence is created by humans, and usually only one aim is pursued when solving applied problems – namely minimizing the so-called loss function, meaning that suitable patterns are searched for among sets of numbers and characters which conceal some real physical meaning. It is clear that such AI is extremely rational, but does not in any way consider the humanistic features of decision-making, and if used in an inexpert or unethical way, can cause harm to humans.

At the same time, it is evident that the objectives of AI design are based on the comprehensive analysis of the problem situation with maximum involvement (ideally of all) of those participating in the problem situation, which requires a full analysis of the system and the formalization of the problem-containing and problem-solving systems. Clearly, in order to take into account the views of the majority, decisions are made on the basis of a large number of criteria, which results in a complex analytical and mathematical task. When interacting with a human, AI should consider not only formal cause-and-effect relations, extracted as patterns in the data, but also the emotional aspect of human relations to the problem situation. AI should be capable of explaining its decisions in such a way that can be readily interpreted by a human (the transition to a so-called “grey box” or “black box”). Finally, when creating AI, it is necessary to exclude incompetence, prejudice and, above all, malicious intentions on the part of the AI developers, which can be achieved with AI automated design based on interpretability, objectivity, responsibility and being human- and socially-oriented. Such automated design technologies could in the future become the de facto or even de jure standard for developers of specific AI solutions.

The International Initiative SIBIASA in the Creation of AI that is Interpretable, Objective, Responsible and Human- and Socially-Oriented

Applied systems analysis (ASA) suggests a holistic (integral) and creative approach to solving problems of an arbitrary nature, which is based on systemic reasoning and methods of gradual transition from weakly structured to partially or completely formalized models of the problem-containing and problem-solving systems. ASA assumes that all problem participants will be involved in the process of working on the problem: the bearer of the problem and all (or most) of the stakeholders, including those closest to the problem bearer and other outside parties. ASA also recommends involving the “opinion” of previous and future generations, and also such “silent witnesses” as the environment. The use of ACA in AI system design permits formal criteria (such as accuracy or mean squared error) to be left to one side in favour of less formal requirements including humanistic or ethical ones.

The SibIASA team has a wealth of experience in the application of original approaches to automatization (without or with minimal involvement of AI specialists), design and control using AI methods and models thanks to self-configuration and self-tuning depending on the task being solved. Such an approach makes it possible to obtain an effective AI model that does not depend on the knowledge, experience or preferences of the developer, and thus, significantly increases the objectivity of the task solution by ASA. This approach is today actively implemented in both the design of human-machine interfaces for the identification and use of individual features of stakeholders (such as gender, age, language and education level) and the emotional component of interaction.

The international initiative SibIASA is linked to the active involvement of young researchers from various countries and cultures and with various languages in solving the problem of creating “humane” AI. Firstly, this is connected to the increasingly popular ASA methodology for the analysis and choice of methods for creating AI. Secondly, it concerns the development of methodology concerning AI automated design. Finally, engaging different specialists from various countries should help in many ways approach the problem of the analysis and use of human- and socially-oriented requirements in AI creation.

At present, SibIASA is collaborating with young and successful researchers from Austria, Germany, Finland, Japan, Slovenia, China, Brazil, the United Kingdom and Nigeria.

---

Currently, one of the leading specialists and the initiator of an international partnership in the field of humanizing AI is Shakhnaz Akhmedova. In 2020, Shakhnaz became a participant in the all-Russian competition “Leaders in international cooperation”.

Line of Development in International Cooperation

The project is intended for the German-speaking countries of Western Europe (Austria, Germany and Switzerland) and channelled towards the creation of a network structure for young scientists and researchers involved in developing artificial intelligence systems.

The choice of the given countries is attributed to a number of factors, among which it is worth highlighting the following:

1. The productive collaboration of SibIASA with universities in these countries leading to the comprehensive support of young scientists through clearly constructed exchange systems, work placements and two-tier research supervision, all of which allows new growth points to be created along with centres of excellence in promising fields of AI.
2. It is in these countries that directions associated with fundamental and applied research in AI are most developed, and in which the greatest successes have been achieved in the development and realization of technologies and systems based on methods of machine learning and AI.
3. It is with these countries that the Russian Federation has established the friendliest ties and closest economic cooperation, the development of which has seen a special emphasis placed on aspects of interaction in the development of modern technologies relating to the humanities.

Initiatives in the Development of Youth Collaboration

The key objective of the project lies in attracting young scientists in the field of AI to actively participate in promoting ideas relating to the humanizing of AI systems. This should be executed in a constructive way, specifically by intensively developing the ideology, methodology and technology behind the automated design of intelligent systems, which are from the beginning targeted to adhere to human values while they analyse the situation, make decisions and perform actions.

In order to achieve the given objective, a project will be realized within SibIASA that will entail the forming of a team of young scientists from the Russian Federation, Austria, Germany and Finland. Their main task will involve attracting young people engaged in science from all countries and fields of science to participate in special workshops dedicated to the humanistic aspects of AI system design at the following conferences:

1. International conference for young scientists ‘Topical Issues in Aviation and Cosmonautics’, SibSU named after M. F. Reshetnev, Krasnoyarsk, April 2021;

2. International applied research conference dedicated to the memory of Chief Designer of Space and Missile Systems, academic Mikhail Fedorovich Reshetnev ‘Reshetnev Readings’, SibSU named after M. F. Reshetnev, Krasnoyarsk, November 2021;

3. ‘International Workshop on Mathematical Models and their Applications’ (IWMMA), SibIASA – University of Ulm, Krasnoyarsk, November 2021.

In the future, similar scientific events are expected at the partner universities in Austria, Germany and Switzerland.

The result of this project aimed at developing youth cooperation with selected countries is the creation of a counterpart to the European Centre for Youth Initiatives specifically for the automated design of humane systems based on AI and a consortium for training personnel in this field.

Experience of Competition Participants’ Interaction with Overseas Partners

Among the more significant individual achievements it is worth noting the following:

1. Winner of the Google Anita Borg Memorial Scholarship for outstanding achievements in research by women in computer science. Invited to the offices of Google in Zurich (Switzerland) and Moscow (Russian Federation) with the aim of holding seminars and presenting project work, 2014.

2. Scholarship from the German Academic Exchange Service (DAAD) for participation in the Euler Program in 2013-14 (placement at the University of Ulm, Germany), and exchange education as part of the International Computer Engineering/Science Program (ICEP) at Ulm University of Applied Sciences (Ulm, Germany) in the spring term of the 2012-13 academic year.

Additionally in the SibIASA team is a participant from the open Russian and German competition ‘Russia and Germany: Scientific and Educational Bridges’, 2020. 

Experience of Social and Scientific Work from Competition Participants

Organization of international conferences:

1. Chairperson of the workshop ‘Optimization Algorithms’ at the ‘15th International Conference on Informatics in Control, Automation and Robotics’ (Porto, Portugal, 29-31 July 2018);
2. Member of the programme committee at the ‘9th International Congress on Advanced Applied Informatics’ (1-15 September 2020, Tokio, Japan)
3. Holding scientific seminars at the offices of Google (2013 – Zurich, Switzerland; 2014 – Moscow, Russian Federation) 
4. Conducting lessons in English for overseas students (Afghanistan, Nigeria, Somalia) at SibSU (mathematical modelling, optimization methods)
5. Conducting lessons for students at the Aichi Prefectural University, Nagakute, Japan (artificial intelligence)
6. Master’s thesis research supervision for the student Chiori Miyajima (Aichi Prefectural University, Nagakute, Japan) in the 2018-2019 academic year.

The Team of Young Scientists:

1. Shakhnaz Akhmedova – PhD, Associate Professor of Higher Mathematics Dep. in Institute of Informatics and Telecommunications, Reshetnev University, Krasnoyarsk, Russia
2. Vladimir Stanovov – PhD, Associate Professor of Higher Mathematics Dep. in Institute of Informatics and Telecommunications, Reshetnev University, Krasnoyarsk, Russia
3. Maria Semenkina – PhD, Research Associate at Heuristic and Evolutionary Algorithms Laboratory, University of Applied Sciences Upper Austria in Hagenberg, Austria
4. Olga Semenkina – System Analyst of the Office of Analytics and Modeling of JSC “Krastsvetmet”, Krasnoyarsk, Russia
5. Anastasiya Polyakova – PhD, Assistant Professor of Dep. of System Analysis and Operations Research, Reshetnev University, Krasnoyarsk, Russia
6. Christina Brester – PhD, Researcher in Dep. of Environmental and Biological Sciences, University of Eastern Finland (UEF), Kuopio, Finland
7. Ivan Ryzhikov – PhD, Researcher in Dep. of Environmental and Biological Sciences, University of Eastern Finland (UEF), Kuopio, Finland
8. Aleksei Vakhnin – PhD student, Research Assistant at Heuristic and Evolutionary Algorithms Laboratory, University of Applied Sciences Upper Austria in Hagenberg, Austria
9. Alina Skorokhod – PhD student, Research Assistant at the Institute of Neuroinformatics, Ulm University, Ulm, Germany
10. Anastasiya Spirina – Dr.rer.nat., System Analyst of the Office of Analytics and Modeling of JSC “Krastsvetmet”, Krasnoyarsk, Russia
11. Tatiana Karaseva – PhD student, Reshetnev University, Krasnoyarsk, Russia
12. Sergei Mitrofanov – PhD student, Reshetnev University, Krasnoyarsk, Russia

Press about the participants of the competition (in Russian):

Implementation Plan of the Team within the Scope of the Project for 2021:

1. Workshop at the international applied research conference ‘Topical Issues in Aviation and Cosmonautics’, SibSU named after M. F. Reshetnev, Krasnoyarsk;

2. Workshop at the international applied research conference dedicated to the memory of Chief Designer of Space and Missile Systems, academic Mikhail Fedorovich Reshetnev ‘Reshetnev Readings’, SibSU named after M. F. Reshetnev, Krasnoyarsk;

3. Workshop at the international scientific conference ‘International Workshop on Mathematical Models and their Applications’ (IWMMA), SibSU named after M. F. Reshetnev, Krasnoyarsk.