【人工智能】研究估計：美國近兩成勞動力或至少一半工作項目受影響 Research suggests around 19% of U.S. workers may see at least 50% of their tasks impacted

人工智能正在改變我們的工作和生活方式，其中以生成式人工智能（Generative AI）帶來的影響尤為顯著。OpenAI、OpenResearch和賓夕法尼亞大學共同發表報告，研究生成式人工智能對各行各業的影響，以及人工智能應用在未來工作場所的潛在可能性。

在研究當中，研究人員採樣人工智能在日常工作上的應用情況，而有關「應用」的情況是否有效，取決於能否將特定工作所需的時間縮減至少50%。值得注意的是，有關的職業只是受生成式人工智能所影響，而非被其取代。

研究評估了生成預訓練變換器（Generative Pre-trained Transformer，GPT）模型和相關技術對各行各業的潛在影響，以及GPT的可應用程度。據研究估計，美國約80%的勞動人口中，至少有10%的工作項目會受GPT影響，約19%勞動人口的一半工作將受影響。

另外，研究人員就GPT-4和大型語言模型（Large Language Models，LLMs）進行研究，結果顯示，編程和寫作相關的工作較受生成式人工智能影響，而涉及科學和批判性思維的職業相對影響較少。預計受AI影響的職業包括：

• 作家
• 數學家
• 報稅員
• 動物科學家
• 公共關係專家
• 傳譯員、筆譯員
• 調查研究員
• 詩人、作詞人、創意作家
• 量化金融分析師
• 網絡和電子界面設計師

研究強調，大多數職業在某種程度上都受GPT影響，其中較高薪的職業將應用更多人工智能。雖然人工智能的「充分應用」並非意味著完全的自動化，但它可以協助人類節省大量時間。在很多無法想像的領域中，生成式人工智能正在重塑我們的職場。儘管一些職業可能將會消失，但一些能透過利用AI改善人們生活的職業，相信可以在2020年以後的十年蓬勃發展。

Artificial intelligence is changing the way we work and live, with the impact of generative AI being particularly significant. A collaborative research by OpenAI, OpenResearch, and the University of Pennsylvania explores the influence of generative artificial intelligence on various industries, revealing the potential possibilities of AI in the future workplace.

In a paper authored by researchers from OpenAI, OpenResearch, and the University of Pennsylvania, the "exposure" of work tasks to AI is examined. The researchers define "exposure" as a measure of whether a GPT system could reduce the time taken for a human to perform a specific work task by at least 50%. It is crucial to note that affected occupations will be significantly influenced or augmented by generative AI, not replaced.

The research assesses the potential implications of GPT (Generative Pre-trained Transformer) models and related technologies on occupations, evaluating their exposure to GPT capabilities. According to the study's estimates, approximately 80% of the U.S. workforce could have at least 10% of their work tasks affected by GPTs, while around 19% of workers may see at least 50% of their tasks impacted.

The researchers base their study on GPT-4 and use the terms large language models (LLMs) and GPTs interchangeably. Their findings suggest that programming and writing skills are more likely to be influenced by generative AI, whereas occupations or tasks involving science and critical thinking skills are less likely to be affected. Occupations experiencing or expected to experience a high degree of AI-based influence and augmentation include:

• Writers, authors
• Mathematicians
• Tax preparers
• Animal scientists
• Public relations specialists
• Interpreters and translators
• Survey researchers
• Poets, lyricists, creative writers
• Financial quantitative analysts
• Web and digital interface designers

The study highlights that most occupations exhibit some degree of exposure to GPTs, with higher-wage jobs generally having more tasks with high exposure. Though full exposure doesn't imply full automation, GPTs could significantly save workers time in completing a large share of their tasks. Generative AI is reshaping the workplace in unimaginable ways. While some occupations may eventually disappear, those that harness AI's productivity and power to create innovations and services improving lives will thrive in the economy of the mid-to-late 2020s and beyond.