The pursuit of artificial general intelligence (AGI) continues to be a key focus in the current AI arms race. AGI is a form of machine intelligence that can solve a wide range of human tasks, adapt to new situations, and exhibit qualities like curiosity, imagination, and intuition. The timeline for achieving AGI varies widely among leading experts in the field, ranging from the next few years to several decades. **Demis Hassabis, Google DeepMind CEO:** Demis Hassabis, CEO of Google DeepMind and a recent Nobel laureate, is optimistic about the near-term arrival of AGI. On the April 20 episode of 60 Minutes, Hassabis predicted that AGI will arrive "in the next five to ten years." By 2030, he anticipates a system that can deeply understand its environment and integrate seamlessly into daily life. However, Hassabis acknowledges that current AI systems lack true curiosity, imagination, and the ability to form new hypotheses independently. These are key elements that AGI must possess to be considered fully realized, and he believes they will emerge within the next five to ten years. **Sam Altman, OpenAI CEO:** Sam Altman, CEO of OpenAI, shares a similar optimism. On the Y Combinator podcast, Altman expressed excitement about making significant progress toward AGI by 2025. OpenAI's advancements in natural language processing (NLP) and generative models have been pivotal, and Altman sees these technologies as stepping stones toward AGI. His confidence is bolstered by the rapid developments in AI research and the company's ambitious goals. **Miles Brundage, Former OpenAI Head of AGI Readiness:** Miles Brundage, who previously headed AGI readiness at OpenAI and left in August, believes we will see some form of AGI within the next few years. Brundage posits that AI systems will soon be capable of performing almost any task a person can do remotely on a computer, such as navigating user interfaces, typing, and even engaging in video chats. This prediction underscores the belief that AGI could manifest in highly practical and versatile forms before we recognize it fully. **Dario Amodei, Anthropic CEO:** Dario Amodei, CEO of Anthropic, an OpenAI competitor, is also confident about the near-future arrival of AGI, predicting it could be achieved by 2026. In an October essay on Anthropoc's website, Amodei described AGI, or what he calls "powerful AI," as being smarter than Nobel Prize winners in multiple fields, multimodal, independent, fast, and cooperative. He envisions AGI as a collection of highly intelligent systems that can be deployed across various domains, likening it to a "country of geniuses in a data center." **Geoffrey Hinton, AI Pioneer:** Geoffrey Hinton, often referred to as the "godfather of AI," is more cautious. Hinton has not provided a specific timeline but强调了AGI需要具备的特质，如深度理解、独立思考和创造性。他指出，当前的AI系统在这些方面仍然存在明显不足。 **Andrew Ng, AI Researcher:** Andrew Ng,另一位知名AI研究者，提出了AGI的两种定义。在一种经济定义下，他认为当80%的工作被AI自动化时，就可以称之为AGI。Ng预测这将在三到五年内实现。然而，如果将AGI定义为具有与最理性的人相同的视觉、运动、语言和逻辑智能，能够像人类一样学习，那么时间线可能会从10年到200年不等。这种宽泛的定义凸显了实现AGI所面临的复杂性和不确定性。 **Richard Socher, You.com CEO:** Richard Socher，前Salesforce高管，现任You.com CEO，同样提供了两种AGI的定义。他认为，当AI能够替代80%的工作时，可以称其为AGI，这可能在三到五年内实现。如果考虑更广泛的定义，包括与人类相同的多模态智能和复杂任务处理能力，那么实现AGI的时间线可能需要更长，从10年到200年不等。Socher的预测强调了AGI实现过程中技术和社会因素的多样性。 **Yann LeCun, Meta AI Scientist:** Yann LeCun，Meta的首席AI科学家，在一月份的世界经济论坛达沃斯年会上表示，AGI不会在短时间内实现。他认为，AGI“不在拐角处”并且可能需要数年甚至数十年。LeCun强调，AGI的实现不会是一蹴而就的，而是一个渐进的过程。在Lex Fridman的播客中，他进一步解释说，AGI的出现不会像科幻电影中那样突然，而是会逐渐发展出来，意味着我们可能会在多个方面逐步看到接近AGI的系统。 **专家们的评估:** 尽管专家们的预测存在差异，但大多数人都同意AGI的发展将是一个长期而复杂的过程。行业内部人士认为，当前的AI技术已经在特定任务上取得了显著进展，但要达到AGI的全面能力，还需要克服诸如好奇心、直觉和创造力等挑战。Hassabis和Altman的乐观态度基于近期技术的快速发展，而LeCun和Hinton则更加谨慎，强调当前技术的局限性和长期研究的必要性。 **公司简介:** - **Google DeepMind:** 成立于2010年，是一家致力于开发通用人工智能的领先研究机构。DeepMind在棋类游戏、蛋白质结构预测等领域的突破，展现了其强大的技术实力。 - **OpenAI:** 成立于2015年，目标是确保AI的安全和发展。OpenAI开发了诸如ChatGPT和DALL-E等知名AI产品，一直在推动AI技术的前沿。 - **Anthropic:** 成立于2021年，专注于开发有道德责任的通用人工智能。Amodei领导的团队在多模态AI和语言处理方面取得了重要进展。 - **Meta AI:** 作为Meta内部的AI研究部门，Meta AI致力于开发各种AI技术。Yann LeCun是该部门的领军人物，他在深度学习和神经网络领域做出了开创性贡献。 总之，虽然AGI的实现时间和具体形式还有很大的不确定性，但众多专家和公司的积极投入表明这一目标正在逐步接近。无论是短期的乐观预测还是长期的谨慎观点，AI的发展都将极大地影响我们的未来。