Nature重磅：图灵测试已死，AI已具备人类水平智能，这一天终于来了

Core Viewpoint - The article emphasizes that Artificial General Intelligence (AGI) has already been achieved, challenging the prevailing skepticism among experts and highlighting the need to recognize the reality of AGI's existence [4][8][34]. Group 1: AGI Development and Recognition - A recent article in Nature declares that AGI has been born, marking a significant milestone in AI development [4][8]. - In a Turing test, GPT-4.5 was identified as "human" by 73% of participants, indicating that AI has surpassed human performance in mimicking human behavior [5][6]. - Despite evidence of AI's capabilities, 76% of top AI researchers still believe AGI is far off, suggesting a disconnect between reality and perception [7][15]. Group 2: Levels of AI Intelligence - The article outlines a three-level framework for assessing AGI: - Level 1: Turing Test level, which has been passed by current AI [10]. - Level 2: Expert level, where AI can solve complex problems and generate scientifically valid hypotheses [11][12]. - Level 3: Superhuman level, which AI is approaching but has not fully achieved yet [13]. Group 3: Misconceptions About AGI - Many misconceptions about AGI stem from unrealistic expectations, such as the belief that AGI must be all-knowing or think like humans [16][18]. - The article argues that true intelligence should be measured by breadth and depth across various fields, not by perfection in every domain [17]. - It challenges the notion that intelligence requires a physical form or emotional capacity, emphasizing that problem-solving ability is the core of intelligence [18]. Group 4: Debunking Common Critiques - The article systematically addresses popular critiques of AI, such as the idea that AI is merely a "Stochastic Parrot" and lacks true understanding [21]. - It refutes the claim that AI's lack of physical presence limits its intelligence, citing examples of individuals like Stephen Hawking who maintained intelligence despite physical limitations [22]. - The article also counters the argument that AI's learning speed is inadequate compared to humans, highlighting that efficiency does not equate to capability [24][25]. Group 5: Implications of AGI's Existence - The recognition of AGI's existence prompts a reevaluation of societal and ethical considerations, including coexistence, responsibility, and governance [34]. - The article concludes that humanity is no longer the sole intelligent entity on Earth, urging a shift in perspective towards this new reality [34][34].