UN report says policymakers are struggling to keep up with pace of AI development

Artificial intelligence development has been progressing at such a rapid pace that current governance systems are unable to keep up, the UN's Independent International Scientific Panel on Artificial Intelligence says in its preliminary report. The panel, consisting of members from around the world, will provide the information needed to stage the UN Global Dialogue on AI Governance. It will take place in Geneva, where member states will discuss how to manage the technology, and is scheduled to begin on July 6. 

In its report, the panel discusses how quickly AI capabilities have evolved over the past few years. Apparently, the complexity of tasks AI models can accomplish has been doubling every few months. The report admits that AI has massive benefits for humanity, including accelerating drug discovery and vaccine development and contributing greatly to antibiotic resistance research. Doctors can also use AI systems for early detection of illnesses, such as breast cancer, and scientists can use AI as early warning systems for food insecurity. 

However, the report also expounds on the new kinds of harms AI systems have created and can create. People have been using AI to generate and distribute sexually explicit deepfakes, including child sexual abuse materials, of real people. California launched an investigation into Grok back in January over nonconsensual deepfakes and CSAM, for instance. AI can also generate false information that appears true, and criminals can use AI systems to aid their cyberattacks. 

Some AI models can be such sycophants; they reinforce a user's harmful behaviors, which could even lead to suicide, the report says. The panel also warns that as AI models become more autonomous, it becomes harder to monitor and control them. And then there's the growing concern over massive data center buildouts to power AI systems, which could harm the communities in their vicinities.

The UN panel explains that policymakers have struggled to keep pace with AI development because current governance systems were not designed for technologies that evolve so quickly. Typically, authorities need scientific data before introducing regulations, but by the time there's enough to understand the technology better, AI systems may already have moved on. 

"The report finds that stronger independent evaluation, international cooperation and common standards are needed to ensure AI systems remain safe, transparent and accountable," the panel writes. Without proper safeguards in place, AI technologies "could deepen inequality, spread misinformation, threaten human rights, disrupt labor markets" and could be a powerful tool "few governments and companies" have access to.

The report notes that access to AI systems is heavily concentrated in developed countries and that most systems are developed in the US and China. Most developing countries lack the infrastructure and expertise needed to benefit from AI truly. "The challenge," the panel says, "is finding a way to unlock AI's enormous benefits while preventing its growing risks." The UN panel, whose role is scientific rather than regulatory, will continue assessing AI technologies that authorities can use to develop policy. It's expected to publish a more comprehensive report next year.