Joseph Sifakis, the Turing Award laureate, expressed at the China Development High-Level Forum 2024 Annual Meeting on March 24th, "China can leverage its vast industrial infrastructure to take a leading position in the journey towards autonomous artificial intelligence."
Sifakis, honored with the Turing Award in 2007 for his contributions to model-checking theory and its applications, highlighted during the conference that human society is still in the era of "weak artificial intelligence." The utilization of AI systems primarily revolves around three functions: assistants, monitors, and controllers. For the latter two functions, there is a requirement for systems engineering that integrates generative artificial intelligence with traditional information and communication technologies. Furthermore, it necessitates new foundations and technologies to ensure security and reliability. China, with its extensive industrial system, has the potential to excel in the field of artificial intelligence.
On March 24th, Joseph Sifakis, the Turing Award laureate, delivered a speech at the China Development High-Level Forum 2024 Annual Meeting.
"Artificial intelligence itself is neither good nor bad," he said, emphasizing the importance of wise utilization and risk prevention through regulation.
Sifakis went on to outline several key risks facing the development of artificial intelligence. One is technical risk, and another is human risk, stemming from misuse or unintended consequences of AI, which can be mitigated through regulation or legal frameworks.
Furthermore, Sifakis identified two categories of societal risks: one concerning the balance between choice and performance. "If we cannot ensure that systems use reliable information in a fair and neutral manner, decision-making authority should not be delegated to the system," he emphasized. The second is the enhancement of performance, which needs to be balanced against the loss of human control.
Regarding regulation, Sifakis highlighted differences between the European Union and the United States. He noted that the EU has stricter and more comprehensive AI regulations, requiring high reliability of AI systems in critical applications, thus excluding many "unacceptably high-risk" applications. In contrast, US regulation is comparatively less mandatory, "not based on very strict technical standards for assessing AI systems," he remarked.
"While there is recognition of the need to regulate artificial intelligence, there is currently no consensus on how to do so effectively," Sifakis concluded.
