AI Progress and Governance Debated Globally
Seventy years after the term was coined, Dartmouth hosted a gathering to assess the future of artificial intelligence. The event coincides with the release of new benchmarks that have pushed advanced AI systems to new heights in reasoning and problem-solving. In response to the rapid progress, the United Nations has announced the creation of an international scientific panel to analyze AI's impacts and advise on risk management.
The original 1956 Dartmouth Summer Research Project on Artificial Intelligence involved just 10 researchers, including pioneers John McCarthy, Marvin Minsky, and Claude Shannon. Their proposal aimed to explore the conjecture that "every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it." Recent AI progress is often measured by performance on complex benchmarks designed to test knowledge and reasoning. One such test is the Massive Multitask Language Understanding (MMLU), which covers 57 academic subjects like math, physics, and law. OpenAI’s GPT-4.1 model has achieved a score of 90.2% on this benchmark, a significant jump from earlier models that performed at near-random chance levels. Other, newer benchmarks are designed to be even more challenging as AI capabilities advance. Tests like GPQA (Graduate-Level Google-Proof Q&A) and SWE-bench for coding are pushing the limits of current systems. For instance, on the difficult "Diamond tier" of the GPQA benchmark, Google's Gemini 2.5 has shown a strong performance edge. In response to these rapid advancements, the UN Secretary-General has convened a High-level Advisory Body on AI. This group consists of 39 experts from 33 countries, drawn from government, the private sector, and academia to ensure a globally inclusive approach. The advisory body's mandate is to analyze and advance recommendations for the international governance of AI. Its work is intended to feed into negotiations for a proposed Global Digital Compact and support preparations for the 2024 Summit of the Future. The group includes prominent figures such as OpenAI CTO Mira Murati.
