Former OpenAI board member Helen Toner emphasized the necessity of implementing a reporting mechanism for incidents involving artificial intelligence (AI) failures, akin to the protocol in place for airplane crashes, during a TED conference talk on Tuesday (April 16).
Toner, currently serving as a director at Georgetown University’s Center for Security and Emerging Technology, underscored the importance of transparency and accountability within AI companies. She stressed that these entities should disclose details about their projects, capabilities, and risk management strategies, as reported by Bloomberg.
The call for transparency extends to the auditing of shared information, Toner suggested, emphasizing the need for independent verification to ensure accuracy and reliability beyond self-assessment by AI companies.
Highlighting potential AI-related issues, Toner referenced the threat of AI-enabled cyberattacks as an example of the technology’s potential adverse impact.
With eight years of experience focusing on policy and governance surrounding AI, Toner provided insights into the efforts of both government and industry sectors to navigate the challenges posed by AI deployment. She emphasized the significance of comprehensive regulation and oversight in managing AI’s evolution, as per Bloomberg’s report.
In a previous interview with CNBC in June 2023, Toner addressed debates within the industry regarding the establishment of a dedicated regulatory agency for AI. She questioned whether existing sector-specific regulatory bodies were adequate or if a centralized authority for all AI-related matters was necessary.
In another development, the United States and the United Kingdom announced a collaborative effort in early April to formulate safety assessments for advanced AI technologies. This initiative aims to harmonize scientific approaches between the two nations and expedite the creation of robust evaluation methodologies for AI models, systems, and agents.