How AI Will Answer Questions We Haven’t Thought to Ask | Aravind Srinivas | TED

这位演讲者，一位毕业于伯克利大学的人工智能博士，也是 Perplexity 的创始人，探讨了一个有趣的趋势：越来越多的学者开始创立科技公司，尤其是在人工智能领域。他注意到，在人工智能会议的演讲者中，博士学位的比例很高，但与之形成对比的是，一项研究显示，过去二十年来，美国博士创立的初创公司数量有所下降。这一观察促使他思考上一个由学者创立的主要科技公司，并认为谷歌就是这样一个例子。然后，他幽默地将这种情况与人们经常指责 Perplexity 试图“杀死谷歌”进行类比，强调他们的重点在于建设和创新。 演讲者接着提到了 2000 年对拉里·佩奇的采访，佩奇设想人工智能是谷歌的终极版本，能够理解和回答网络上的任何问题。这一几十年前的愿景准确地预测了搜索的未来轨迹。他以此来引出一个核心问题：虽然技术往往是关于预测搜索或人工智能的未来，但同时考虑人类的未来至关重要。 演讲者认为，人工智能最终将提升我们的人类体验，并强调坚持不懈地提问的重要性，这既是学术界的核心原则，也是苏格拉底式的方法。他用一个幽默的轶事来阐释这一点，即南印度父母质疑放弃稳定工作而加入初创公司的决定。此外，他还提到了大卫·多伊奇关于人类好奇心延伸到熟悉事物的概念，即使在拥有现有知识的情况下，也总是寻求新的问题。他将这种坚持不懈的提问与 Perplexity 的起源联系起来。 他回顾了 Perplexity 如何在他从学术界转型为企业家期间，因个人对传统搜索引擎的沮丧而诞生。面对海量的信息和有限的时间，他和他的联合创始人创建了一个由人工智能驱动的 Slack 机器人，可以通过网络上精心研究的答案来回答问题。这个机器人迅速发展壮大，为用户提供带有引用的可信答案，镜像了学术界引用来源的做法。这种信任至关重要，可以促进更好的后续问题并培养更深入的知识。 演讲者强调了知识的可访问性对人类进步的深远影响。历史上，获得最佳答案的机会仅限于那些拥有资源和人脉的人。然而，人工智能驱动的搜索正在使信息获取民主化，从而拉平了哈佛教授和发展中国家的学生之间的差距。这种转变将研究的边际成本降低到接近于零。 在这个新时代，获取知识不再是障碍，而提出问题的质量将成为决定性因素。演讲者认为，只要坚持不懈地提问，人类探索的潜力是无限的。他用婴儿与生俱来的好奇心，以及人工智能生成的答案所产生的看似无穷无尽的问题分支来说明这一点。 最终，演讲者认为，技术和人工智能的未来取决于人类好奇心的未来。他相信，随着人工智能越来越擅长回答问题，人类的好奇心和坚持不懈地提问的品质将变得更加重要。我们开发用来提问的工具以及我们用答案创造的创新将定义技术的未来，更重要的是，人类的未来。他以对未来问题的提出和因我们集体好奇心而取得的进步充满希望的期待结束了演讲。

The speaker, a PhD in AI from Berkeley and founder of Perplexity, explores the interesting trend of academics increasingly founding technology startups, particularly within the AI field. He notes the prevalence of PhDs among AI conference speakers and contrasts this with a study showing a decline in startup formation by US PhDs over the past two decades. This observation leads him to ponder the last major technology company founded by academics, identifying Google as the example. He then draws a humorous parallel with the frequent accusation that Perplexity is trying to "kill Google," emphasizing instead their focus on building and innovation. The speaker then references a 2000 interview with Larry Page, who envisioned AI as the ultimate version of Google, capable of understanding and answering any question on the web. This vision, made decades ago, accurately predicted the future trajectory of search. He uses this to frame a central question: while technology is often about anticipating the future of search or AI, it’s crucial to consider the future of humanity alongside technological advancements. The speaker posits that AI will ultimately enhance our human experience, emphasizing the importance of relentless questioning, a core principle of both academia and the Socratic method. He illustrates this with the humorous anecdote about South Indian parents questioning a decision to leave a stable job for a startup. Furthermore, he references David Deutsch's concept of human curiosity extending to the familiar, always seeking new questions even with existing knowledge. He ties this relentless questioning to the origins of Perplexity. He recounts how Perplexity emerged from his personal frustration with traditional search engines during his transition from academia to entrepreneurship. Faced with an overwhelming amount of information and limited time, he and his co-founders created a Slack bot powered by AI to answer questions with well-researched responses drawn from the web. This bot quickly evolved into something much bigger, providing users with trusted answers accompanied by citations, mirroring the academic practice of source attribution. This trust is paramount, enabling better follow-up questions and fostering deeper knowledge. The speaker emphasizes the profound impact of accessible knowledge on human progress. Historically, access to the best answers was limited to those with resources and connections. However, AI-powered search is democratizing access to information, leveling the playing field between a Harvard professor and a student in a developing nation. This shift reduces the marginal cost of research to near zero. In this new era, access to knowledge becomes less of a barrier, and the quality of questions asked becomes the defining factor. The speaker believes that the potential for human inquiry is infinite, as long as relentless questioning persists. He illustrates this with the innate curiosity of babies, and the seemingly endless branching of questions that arise from AI-generated answers. Ultimately, the speaker argues that the future of technology and AI hinges on the future of human curiosity. He believes that as AI becomes increasingly adept at answering questions, the human quality of curiosity and relentless questioning will become even more critical. The tools we develop to ask questions and the innovations we create with the answers will define the future of technology and, more importantly, the future of humanity. He concludes with a hopeful anticipation of the questions that will be asked and the progress that will be achieved as a result of our collective curiosity.