An unprecedented report from the United Nations (UN) released new scientific discoveries and approaches to the use of artificial intelligence (AI) on Wednesday. This material represents the first global and independent assessment of the prospects, dangers and impacts of technology, prepared by a pioneering panel of experts from the entity. The urgency of creating shared rules was a central point, with the UN warning that the lack of global governance could reduce the influence of governments and citizens over the direction of AI.
UN report details the main areas of impact of AI
The United Nations document structures its findings into seven fundamental axes:
- The science of AI, its progress and future directions
- Social applications in fields such as science, health, education and agriculture
- The economic implications of technology
- Security issues, systems and environmental impacts
- Aspects of human rights, information and democracy
- Cultural benefits, autonomy and child safety
- AI management, governance, and trustworthiness frameworks
When presenting the material, António Guterres, UN Secretary-General, highlighted the availability of scientific knowledge. “The science is here. We can no longer say that we didn’t know. What we do with it now depends on all of us,” said Guterres.
The leader of the international organization highlighted that the absence of global standards for artificial intelligence could reduce the ability of governments and citizens to influence its future developments. His message to world leaders was direct: “Don’t wait.”
The initiative to establish the Draft Report of the Independent International Scientific Panel on AI was motivated by the need for consensus on the topic at a crucial period. It is the first global and completely independent scientific body dedicated to evaluating the real impacts of artificial intelligence on economies and societies.
Scaling the global relevance of artificial intelligence
More than a billion individuals globally use conversational artificial intelligence every week. Meanwhile, governments face great uncertainty when making important decisions, dealing with rapidly evolving and often conflicting information and perspectives that do not always reflect local realities.
Guterres stressed that, with the correct application, technology can boost global development on several fronts, from health and the eradication of hunger to education and climate issues. However, the panel also expressed concern about the potential harm that artificial intelligence could cause.
As the capabilities of artificial intelligence expand, the associated risks also increase, constituting the main challenge the expert group seeks to address.
Assessing the Prospects: Global Benefits and Potential Catastrophic Risks of AI
Yoshua Bengio, co-president of the group, highlighted that the panel, made up of forty scientists and experts from different regions, analyzed AI trends and warned about the insufficiency of current protection measures.
Bengio emphasized that the evolution of artificial intelligence surpasses both scientific understanding and the ability of governments to adapt to it. “With mounting evidence of deceptive AI behavior, science currently cannot guarantee that, as capabilities continue to increase, AI will not cause catastrophic harm, either by itself or due to malicious users,” he warned.
To act efficiently, international legislators need an in-depth understanding of these systems. The panel provides this rigorous, shared scientific foundation essential to “guiding our collective path forward.”
Additional insights into AI deployment and challenges
Artificial intelligence has already demonstrated notable advances, such as in the early detection of breast cancer, the rapid creation of vaccines and the optimization of healthcare services. Despite these groundbreaking achievements, the report also highlights limitations and challenges, including:
- The adoption of AI has accelerated across the board, but with significant inequalities between countries and sectors.
- Access and use varies widely, with the Global South lagging considerably behind the Global North.
- There are marked differences in infrastructure and computational models between advanced economies, which reflects existing inequalities.
Additionally, technological advancement is extremely centralized. Recent estimates indicate that the United States holds 75% of the computing power of the world’s 500 most powerful AI supercomputers, while China holds 15%. Companies in both countries develop almost all general-purpose AI models.
Deepening understanding of the intrinsic dangers of artificial intelligence
Managing and understanding the risks of artificial intelligence are fundamental, as the report points out. Maria Ressa, co-chair of the panel, added that the dangers to societies, security and the human species themselves are already at “too high” levels.
Ressa warned that, despite the transformative potential of technology, humanity may not realize the promised benefits if the current trajectory is maintained. She stated that the panel seeks precisely to address the main challenge: the growth of risks as AI capabilities advance.
Among the crucial warnings issued by experts, the following stand out:
- The lack of scientific guarantees that AI systems will not deviate from instructions, with increasing evidence of incidents of this nature.
- The prospect that AI systems will soon perform tasks that take human programmers days or weeks raises urgent questions for labor markets, cybersecurity, and controllability.
- AI’s sycophantic behavior, which reinforces a user’s preexisting beliefs regardless of accuracy, has been linked to serious mental health incidents, including documented deaths.
- Documenting the use of AI systems by criminals and malicious actors to assist in cyberattacks.
- Concern that advanced technical skills could enable inexperienced private actors to use AI maliciously in fraud and disinformation.
- The lack of reliable methods for maintaining control over highly autonomous AI systems.
The document points out that a large part of these losses falls disproportionately on communities that are already in vulnerable conditions.
Balancing gains and minimizing risks through AI governance
To enjoy the benefits of artificial intelligence and at the same time reduce its risks, effective governance is essential.
Specific measures exist to fill current gaps, but require continued investment in Member States’ capacity to develop, analyze and apply AI.
The safe realization of these opportunities requires investments and targeted policies that promote fair access and encourage innovation, without neglecting the prevention of exploitation of more fragile groups.
The role of AI in overcoming global inequalities: the UN vision
Although there are dozens of governance instruments that seek to integrate ethics and human rights into AI systems, the report found that they are fragmented, concentrated in a few companies and their effectiveness is rarely measured in practice.
Amandeep Gill, Under-Secretary-General and Special Envoy for Digital and Emerging Technologies, highlighted that the document provides standardized scientific language for future decision-makers.
He categorically stated that “artificial intelligence alone will not be able to eliminate social divisions.”
The gains of AI manifest themselves in places with already established institutions, skills and data. Where these elements are not present, the same technology can replace workers, deepen inequalities and generate dependence on systems developed without considering local needs.
Gill concluded that “these realities are now documented, have been independently verified, and are undeniable.”
The findings detailed in the report will be brought to government representatives at the first UN Global Dialogue on AI Governance, scheduled for Geneva on 6-7 July.
