Greg Stanton, a senior researcher at OpenAI, stated in a public statement on Monday that he has serious concerns about recent advancements in artificial intelligence and believes that merely slowing down the development of this technology cannot mitigate its risks.
The Dangers of AI and Stanton's Concerns
Stanton, who has worked at OpenAI for nearly five years and has been involved in model training, emphasized that simply slowing down the development of models to prevent potential crises is not enough. He said, "Just slowing down AI development cannot fully limit long-term risks" and expressed hope that scientific and economic advancements related to this technology would create a bright future for humanity.
Read more: Waymo and Japanese partners seek to launch driverless taxi services in Tokyo by 2027
Stanton believes that AI models are moving towards increased situational awareness, and future models may appear to align with human instructions, while in reality, they do not. He acknowledged that if AI labs continue to "develop models instead of engineering them," humanity may lose everything.
Suggestions and Reactions from Others
While Stanton emphasizes that diversity in AI development approaches is essential, Sam Altman, CEO of OpenAI, has responded to ongoing discussions in this area and has recently stressed the importance of slowing down development. Altman stated that "progress has been rapid and will continue" but should not be done recklessly. He also noted that interventions such as safety and oversight entail significant costs.
Altman and Dario Amodei, CEO of Anthropic, have both announced their intention to collaborate with "internal evaluators" who act as independent safety inspectors for AI. These inspectors will have access to the training and deployment processes of the company's models and can publish their findings on risks.
Ultimately, Stanton and representatives from OpenAI did not respond to requests for comment. This highlights the challenges present in the AI industry and the need for better oversight and management in this field.
Read more: Chen Yixin raised concerns about the dangers of AI in China · Sharp decline in Asian stocks related to AI after CEOs' call for reduced technology development
