Автор: ПОДАРОК ХЕРАМБУ ДЕВИПРАСАД | PODAR HERAMB DEVIPRASAD
A brief overview
With the advent of artificial intelligence systems such as GPT-4 in the last few months, it has become urgent to develop common applicable standards that promote AI security, as well as the use of such tools for economic growth. Given the unprecedented pace of development of AI capabilities, the BRICS countries have a unique opportunity to use these technologies to solve specific problems inherent in a particular region and integrate AI solutions into the ecosystem of government startups. This approach can make a significant contribution to solving pressing socio-economic problems unique to these countries. However, the potential risks associated with AI, especially regarding data privacy and possible abuse, necessitate a comprehensive and collaborative approach to policy formulation.
This work outlines a number of policy recommendations and applicable standards aimed at the safe, ethical and inclusive development and deployment of AI technologies in the BRICS countries. These recommendations cover a number of critical areas, including data protection, algorithm transparency, AI in education, and the establishment of a joint regulatory system. By adopting these standards, the BRICS countries position themselves at the forefront of AI innovation, ensuring that the development of AI is consistent with the collective interests and values of their societies.
Introduction:
Why focus on BRICS?
Brics is an association of five major countries, including Brazil, Russia, India, China and South Africa. The developing economies of these countries cover 30% of global GDP, according to the IMF, and almost 46% of the world's population, according to the United States.
Image from Visualcapitalist.com
Image from Visualcapitalist.com
Moreover, the BRICS countries have a young population, they are spread all over the world and new members will join them soon.
Undoubtedly, these countries are major players in the geopolitical landscape at the regional and international level and have a lot of potential. In this light, other countries have to pay attention to what this group of countries is doing, as it has far-reaching consequences.
The BRICS structure proposed in the recommendations
Why focus on Artificial Intelligence?
At the recent BRICS summit, Chinese Prime Minister Xi Jinping said: “We need to jointly reflect risks and develop AI management systems and standards with a broad consensus to make AI technologies safer, more reliable, controllable and fair.”China“ has set a government goal to invest 150 billion US dollars in AI and become a world leader in this field by 2030. Countries such as the United States, the United Kingdom and the European Union are also moving towards regulation and focusing on using AI for their economies. Recently, we saw how the President of the United States signs an executive order on AI as an issue of national importance, and how the European Union discusses its landmark AI Act.
AI is progressing at breakneck speed and many predict that it will soon reach the level of human intelligence.
As tools using advances in natural language processing penetrate businesses and society, they could cause a 7% increase in global GDP (or almost $7 trillion) and boost productivity growth by 1.5 percentage points over a 10-year period, according to Goldman Sachs.
An analysis of the AI index report on the legislative protocols of 127 countries shows that the number of documents containing “artificial intelligence” that were adopted at the legislative level increased from just 1 in 2016 to 37 in 2022. An analysis of parliamentary AI protocols in 81 countries similarly shows that the mention of AI in global legislative proceedings has increased by about 6.5 times since 2016. (report by the Stanford Institute for Human-Centered Artificial Intelligence, 2023)
2023 Stanford Institute for Human–Centered Artificial Intelligence AI Index Report - Number of AI Publications
It is also noteworthy that investments in AI are now 18 times higher than in 2013, which seems to suggest that the expected economic return is very high.
2023 Index Report on AI by the Stanford Institute for Human-Centered Artificial Intelligence
Forecast
AI Impacts conducted a survey focused on trends in the development of AI, aimed at researchers whose work was published in the proceedings of the Conference on Neural Information Processing Systems (NeurIPS) or the International Conference on Machine Learning (ICML), two leading AI conferences, in 2021. Of the 4,271 researchers who were initially contacted, 738 people responded, which means that the response rate was 17%. These researchers provided their predictions on a number of AI-related topics, including estimated dates of occurrence and the expected impact of AI on various professions (Epoch AI, 2021). It is important to note that the results of this survey may show some degree of sample bias. For example, it is possible that researchers with more aggressive views on the timeline and the impact of AI might be more likely to participate in such surveys.
In another survey of leading researchers conducted by Grace et al. In 2018, researchers suggest that AI will surpass human capabilities in several tasks. They predict that AI will succeed in language translation by 2024, writing essays at the university level by 2026, driving trucks by 2027, and managing retail risks by 2031. Looking further into the future, they predict that AI will write a bestseller by 2049 and will perform surgical operations by 2053. In addition, these researchers estimate a 50% chance that AI will surpass humans in all tasks in the next 45 years and automate all human professions in the next 120 years.
Forecast by leading researchers, Grace et al.,2018
The popular Metaculus collective forecasting website predicts that a weak general intelligent AI system will appear in February 2026.
Forecast on the Metaculus website
Ongoing work on AI in the BRICS countries
The BRICS countries have agreed to launch an AI Research Group from the BRICS Institute for the Study of Future Networks in the near future. The BRICS countries have their own National AI Strategies and data protection agencies, which are vital for the protection of human rights and the safe development of AI. The BRICS countries have some of the youngest people in the world who are skilled in IT and able to contribute to the economy through AI.
Moreover, systems such as the UN Sustainable Development Goals, the G20 Declaration on AI and the OECD Principles on AI guide the BRICS countries by developing their own national policies.
AI Journal Publications, 2023 Index Report on AI by the Stanford Institute for Human-Centered Artificial Intelligence
It is especially noticeable that the BRICS countries India and China are the main leaders in AI Journal Publications, with China accounting for almost 40% of all publications between 2010 and 2021.
The image was generated using DALL E-3
Proposed standards:
Below are the standards and policy recommendations that the BRICS Institute's AI Research Group can put into practice.
1. The principle of AI for the public good
The parties agree that AI technologies will be created, developed and deployed to improve public well-being in accordance with the global Sustainable Development Goals.
Source: United Nations Sustainable Development Goals, ISO/IEC 38507:2017 - Governance of information technology for the organization.
2. Guidance through AI
Each party undertakes to use AI to improve educational systems, ensuring accessibility and cultural adaptability.
Source: UNESCO's Beijing Consensus on Artificial Intelligence and Education, ISO/IEC 23894:2020 - Information technology — Risk management for AI systems.
3. Autonomous AI regulation systems
The parties must ensure that autonomous AI systems operate in an environment of accountability and transparency.
Source: European Commission's Ethics Guidelines for Trustworthy AI, ISO/IEC 42010:2011 - Systems and software engineering — Architecture description.
4. Adoption of universal AI guidelines
The parties should adopt a common set of guidelines for AI, ensuring respect for the principles of human rights, diversity and non-discrimination.
Source: OECD Principles on Artificial Intelligence, ISO/IEC 42001 - Information technology — Management system standards guidance.
5. Human-oriented AI decision-making process
AI decision-making processes will be subject to human control, especially in critical areas affecting human rights.
Source: EU's White Paper on Artificial Intelligence, ISO/IEC 27001:2013 - Information technology — Security techniques — Information security management systems — Requirements.
6. Acceptable risks in AI
Each Party should define standards for assessing and managing the risks associated with AI
Source: IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems, ISO/IEC TR 24028:2020 - Information technology — Artificial intelligence — Overview of trustworthiness in artificial intelligence.
7. Data protection and confidentiality
The Parties commit themselves to strict data protection standards.
Source: General Data Protection Regulation (GDPR), ISO/IEC 29100:2011 - Information technology — Security techniques — Privacy framework.
8. Ban on AI in deadly autonomous weapons
The development and use of AI in lethal autonomous weapons is hereby prohibited.
Source: International Committee of the Red Cross's position on autonomous weapons, IEEE P7000 - Model Process for Addressing Ethical Concerns During System Design.
9. Reducing AI bias
Methods will be created to identify and reduce bias in AI systems, ensuring consistency with human dignity.
Source: UNESCO's Recommendation on the Ethics of Artificial Intelligence, ISO/IEC TR 24029-1:2021 - Information technology — Artificial intelligence — Assessment of the robustness of neural networks.
10. Harmonization of AI standards
The parties will cooperate to harmonize AI standards, facilitating interoperability and sharing best practices.
Source: IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems, ISO/IEC 30141:2018 - Internet of Things (IoT) — Reference architecture.
11. Knowledge sharing in AI
A knowledge exchange platform will be created between the Parties to share achievements and solutions in the field of AI.
Source: World Economic Forum's Centre for the Fourth Industrial Revolution, ISO/IEC TR 24368:2020 - Information technology — Artificial intelligence — Use cases.
12. Explainability and transparency of algorithms
Standards will be applied to ensure the explainability and transparency of AI algorithms.
Source: AI Now Institute's Report on Algorithmic Accountability, ISO/IEC TR 24027:2020 - Information technology — Artificial intelligence — Bias in AI systems and AI-aided decision making.
Policy recommendations:
AI policy recommendations among the BRICS countries
Taking into account and recognizing the need for a cooperative approach in the development and management of Artificial Intelligence (AI) technologies in the BRICS countries and striving to create a system of cooperation, ethical standards and mutual development in the field of AI, the following recommendations are proposed.
Cross-border security tools
The parties will create and put in place infrastructure support for monitoring and verifying AI systems, as well as inclusive datasets to create a thriving AI ecosystem.
This will include mechanisms for regular cross-border data transfer and maintenance of data integrity and security.
2) The mechanism of periodic mutual control
A rotating system of periodic mutual control should be established, in which each Party checks the AI policy of the other country, identifying opportunities for cooperation and offering recommendations for achieving the previously mentioned standards.
3) National standards for data storage and cybersecurity
Each Party agrees to establish a national data storage practice with common cybersecurity standards in accordance with international guidelines, guaranteeing the protection of citizens' data from unauthorized access and hacking.
4) Joint Technical Advisory Committee
A joint committee of technical experts from all member countries will be established to advise Governments on best practices in the field of AI. The Committee will also facilitate the exchange of research, development strategies and ethical considerations in the field of AI, working with the BRICS Business Council.
5) Annual rotation of AI checks among member countries
An annual review system is proposed, where each member country evaluates the AI systems and policies of the other country, ensuring compliance with agreed standards and contributing to continuous improvement.
6) Openness to independent checks
All Member States should commit to transparency by remaining open to independent reviews of their National Strategies and AI applications conducted by recognized international monitoring agencies.
7) AI risk classification scheme
The parties should cooperate to create a comprehensive risk classification scheme for AI systems, considering different levels of impact and complexity.
8) National Institutes of AI Security
Each Party is recommended to establish a National Institute for AI Security. These institutions will provide mechanisms for citizens to report issues related to bias against AI or human rights violations, ensuring accountability in the application of AI.
9) The Expanded Brics Technology Coalition (BET)
It is recommended to create an Expanded Brics Technology Coalition (BRICS Extended Tech Coalition – BET) to involve non-BRICS countries. The BET coalition will facilitate technical cooperation, the introduction of standards and the application of strategies in AI.
SCHEMATIC DIAGRAM:
The scheme of the proposed management structure
Limitations:
Much of this is a cyclical process of agreeing on the details to be discussed and other considerations of each individual member country. Some of the limitations of the standards and suggested policy recommendations are presented below:
1) The BRICS Study Group was formed to monitor the development of AI, not to develop policy recommendations. However, the Group's experts can recommend this for effective monitoring and use of AI technologies by member governments.
2) Data storage and cybersecurity policies should balance global cooperation with national interests and issues of data sovereignty.
4) The creation of new committees and institutions (such as the National Institute for AI Security) may require a significant amount of resources. Clear guidance on financing and resource allocation is needed.
5) The strategies cover wide areas and more specific application strategies may be required for different countries, taking into account their unique socio-political and technological landscape.
The feasibility of harmonizing standards among different countries should be carefully evaluated, especially in areas such as data privacy and AI in weapons.
Conclusion:
AI systems are developing at a breakneck pace, and there must be a framework for such development in order to prevent violations of the social structure. Citizens and regulators need to ensure that AI systems are consistent with human values, respect human rights, and are used to maximize social well-being. The BRICS countries have a key role to play in this growth, and they must set the standards for the possibility of using AI with the maximum possible productivity and solving the pressing problems of developing countries in the middle of the 21st century with the aim of becoming a leader by 2100.
The original (with images and graphs) in the application