Sam Altman, CEO of OpenAI, and Greg Brockman, president of OpenAI, Refutation of Controversy Surrounding Super Alignment Team
Sam Altman, the CEO of OpenAI, and Greg Brockman, the president of OpenAI, have directly refuted the controversy surrounding the dismantling of the Super Alignment team, the department responsible for artificial intelligence (AI) safety and ethics. In a statement on X (formerly Twitter) on the 19th, they emphasized the significance of raising awareness about the dangers and opportunities associated with general artificial intelligence (AGI) to allow the world to prepare for its emergence.
Advancements in AI Safety and Ethics
The duo stated, "We have demonstrated the potential for advancing deep learning and have conducted thorough analyses of its implications. Additionally, we have spearheaded efforts to establish international AGI regulations for the first time and have assessed the potential risks posed by AI systems."
Altman and Brockman highlighted OpenAI's commitment to laying the groundwork for the safe deployment of proficient systems. They acknowledged the challenges of ensuring the safety of new technologies, citing their efforts in the secure release of GPT-4. The continuous refinement of the model's behavior and the enhancement of abuse monitoring reflect the lessons learned during the distribution process.
Challenges Ahead
Despite their accomplishments, Altman and Brockman acknowledged the complexity of preparing for AGI risks, emphasizing the need to intensify safety measures in alignment with the significance of the new model. They dismissed the notion that the superalignment method alone could avert AGI risks, underscoring the absence of established guidelines for navigating the path to AGI. They proposed that empirical insights could offer valuable guidance moving forward.
OpenAI's ongoing efforts aim to accentuate the positive aspects of AI development while mitigating potential hazards. The duo reiterated their commitment to addressing significant risks associated with AI.
Disbandment of Super Alignment Team
The statement was issued following the abrupt disbandment of the Super Alignment team, which was established to oversee the regulation of general AI in advance. The team focused on developing technologies to ensure that super-intelligent AI behaves in a manner aligned with human intentions, aiming to explore methods for controlling autonomous decision-making beyond human capabilities without causing harm.
Internal conflicts arose within the team, particularly concerning the allocation of computing resources. The team's demand for 20% of computing resources, including GPUs, led to concerns about sluggish progress. Subsequently, team members, including prominent scientists such as Ilya Sutskever, Leopold Aschenbrenner, and Pavel Izmailov, departed from the company. Others, like William Saunders, Cullen O'Keefe, and Daniel Kokotajlo, are also believed to have left OpenAI.
In the aftermath, Jan Leike, a former DeepMind researcher, expressed concerns about the implications of creating machines more intelligent than humans. He emphasized OpenAI's responsibility in safeguarding humanity's interests. Despite the challenges, he noted a perceived shift in prioritizing AI safety over the development of popular products in recent years.
