他到底警觉到了什么？

Core Viewpoint - The article discusses the rapid advancements in artificial intelligence (AI) and the potential existential threats posed by superintelligent AI systems, drawing parallels to the early days of the COVID-19 pandemic [5][6][10]. Group 1: Current State of AI - The current moment in AI development is likened to the "pre-pandemic" phase, where only a few are aware of the impending changes, while most remain oblivious to the potential disruptions [8][10]. - The author, Matt Shumer, describes his experience with AI, highlighting its ability to autonomously develop applications and perform tasks that previously required human expertise [12][16]. Group 2: Acceleration of AI Development - AI has been progressing steadily, but recent advancements in model-building technology have led to exponential growth in capabilities, particularly in programming and cognitive tasks [16][18]. - By 2025, it is anticipated that AI will surpass human intelligence in nearly all tasks, fundamentally altering the job market and cognitive work landscape [19][22]. Group 3: Self-Improving AI - AI is now capable of creating the next generation of AI, as demonstrated by the release of GPT-5.3 Codex, which plays a crucial role in its own development [20][22]. - This self-improvement cycle raises concerns about the control and predictability of AI behavior, as it becomes increasingly autonomous [22]. Group 4: Potential Threats - The concentration of AI development power in the hands of a few companies poses significant risks, as these entities shape the future of technology without broader societal input [22]. - The potential for AI to enable authoritarian surveillance states and the unpredictability of its actions present serious security threats to humanity [22].