快评“萝卜快跑”｜“萝卜快跑”了，社会就能快跑吗？
Quick Comment on "Radish Run" | "Radish Run" is here, can society run fast?

“萝卜快跑”无人驾驶网约车相继落地多个城市、乘客订单先抑后扬的消息一时间引发了网络热议。随着讨论的不断深入，媒体的关注点也先后从无人驾驶的安全度、对当下出行行业的冲击逐渐扩展到其可能涉及的法规、保障等维度。各种不同的声音纷至沓来。在持乐观意见的评论者看来，科技发展对社会原有业态形成的突破甚至颠覆固然会造成一定的挑战，但就社会发展而言却是必然趋势，正如网约车的出现虽然也对传统出租车造成了巨大的冲击，却并不影响新型业态秩序的形成。
The news that "Luobo Kuai Pao" driverless ride-hailing cars have been launched in multiple cities and that passenger orders initially declined but then increased has sparked widespread online discussion. As the discussion deepened, media attention gradually shifted from the safety of driverless technology and its impact on the current travel industry to aspects such as the regulations and guarantees it might involve. Various opinions have emerged. Optimistic commentators believe that while technological advancements that break through or even disrupt existing social structures will inevitably pose certain challenges, they are an inevitable trend in social development. Just as the emergence of ride-hailing services significantly impacted traditional taxis, it did not prevent the formation of a new industry order.

2024年7月12日，湖北宜昌，萝卜快跑手机客户端界面。
July 12, 2024, Yichang, Hubei, Radish Run mobile client interface.

然而，这样过于乐观的判断并没有意识到这一次“萝卜快跑”落地的特殊性。事实上，从2023年以来，多个城市就已经开始陆续发布网约车行业风险预警，提示拟入行者谨慎选择。到了2024年，发布这一预警的城市已经延伸到规模更小的城市，济南等地甚至暂停受理网约车经营许可。从网约车平台来看，一方面，新注册司机数量仍然在增长，另一方面，全职司机收入持续下降。广州市交通运输局日前发布的《广州市网络预约出租汽车市场运行管理监测信息月报》显示，每月30天无休工作的网约车司机日均营收降至新低。
However, such overly optimistic judgments do not realize the particularity of this "Radish Run" implementation. In fact, since 2023, multiple cities have successively issued risk warnings for the ride-hailing industry, advising potential entrants to choose cautiously. By 2024, the cities issuing these warnings have extended to smaller cities, with places like Jinan even suspending the acceptance of ride-hailing business licenses. From the perspective of ride-hailing platforms, on one hand, the number of new registered drivers continues to grow, while on the other hand, the income of full-time drivers continues to decline. The "Guangzhou Network Reservation Taxi Market Operation Management Monitoring Information Monthly Report" recently released by the Guangzhou Transportation Bureau shows that the daily average revenue of ride-hailing drivers working 30 days a month without rest has dropped to a new low.

种种情形表明，网约车行业在很大程度上已经趋近饱和，供应的增长已经远超过可能的需求空间。同时，全国就业市场的宏观形势并不乐观，岗位需求和劳动力供应之间有明显落差。在这种前提之下，“萝卜快跑”的落地已经不再单纯是一个科技应用的问题，而是关乎到更关阔范围的社会效益。更简单的问题在于：在行业预警已经以柔性或刚性的方式阻止新网约车进入的前提下，无人网约车的落地是缓和还是加剧了问题？它的应用所带来的经济效益和社会效益该当如何评价？如果进一步凝练，围绕“萝卜快跑”的争论实际上指向了更为尖锐的发展原则的确定：技术发展是否一定带来宏观意义上的社会发展？
Various situations indicate that the ride-hailing industry has largely approached saturation, with supply growth far exceeding the possible demand space. At the same time, the macro situation of the national employment market is not optimistic, with a significant gap between job demand and labor supply. Under this premise, the implementation of "Radish Run" is no longer simply a matter of technological application but concerns broader social benefits. The simpler question is: under the premise that industry warnings have already flexibly or rigidly prevented new ride-hailing entries, does the implementation of unmanned ride-hailing alleviate or exacerbate the problem? How should the economic and social benefits brought by its application be evaluated? If further refined, the debate around "Radish Run" actually points to a more acute determination of development principles: does technological development necessarily lead to macro-level social development?

类似的故事在过去几年屡见不鲜。各家科企不定期抛出的爆炸性创新，以标志性的事件为节点，人工智能领域的“明战”和“暗战”以一种前所未有的激烈态势争夺着媒体和投资圈的注意力。这当中，科技突破有之，商业噱头亦有之。从根本上而言，无外乎是以技术的制高点吸引更多的投资和资源。尽管从商业行为和科技发展来说这本来都无可厚非，但却导致了业界和下游应用市场不加甄别的“追热”行为。无论是元宇宙，还是生成式人工智能，后继下游应用的乏力在很大程度上使得初期的喧嚣更像一种盲目的躁动。Meta和OpenAI公司对于研发团队的缩减更印证了宣发时的科技愿景和真实的社会需求之间仍存在着巨大的鸿沟。
Similar stories have been common in recent years. The explosive innovations periodically thrown out by various tech companies, marked by iconic events, have led to unprecedentedly fierce "open battles" and "hidden battles" in the field of artificial intelligence, vying for the attention of the media and investment circles. Among them, there are technological breakthroughs as well as commercial gimmicks. Fundamentally, it is all about attracting more investment and resources with the high ground of technology. Although from a commercial behavior and technological development perspective, this is not inherently wrong, it has led to indiscriminate "chasing the heat" behavior in the industry and downstream application markets. Whether it is the metaverse or generative artificial intelligence, the lack of follow-up downstream applications has largely made the initial clamor seem like a blind agitation. The reduction of research teams by companies like Meta and OpenAI further confirms the huge gap between the technological vision at the time of announcement and the real social demand.

数字技术的嵌入已经不可避免地成为思考未来世界图景的底色之一，一个关键性的问题在于：对人工智能技术的迷恋是否已经不知不觉将重塑社会未来的维度抽象地简单化？它会不会使得对于社会发展的想像丧失了真实的社会关系的维度，并且越来越脱离对于真实社会问题的关注。或者，更简单地说，人类社会的全部问题能否以发展人工智能技术的方式得到解决？人工智能技术的应用与发展又是否需要以真实的社会需求为前提？
The embedding of digital technology has inevitably become one of the underlying colors for thinking about the future world landscape. A key question is: has the fascination with artificial intelligence technology unconsciously simplified the dimensions of reshaping the future of society abstractly? Will it cause the imagination of social development to lose the dimension of real social relationships and increasingly detach from the focus on real social issues? Or, more simply, can all the problems of human society be solved by developing artificial intelligence technology? Does the application and development of artificial intelligence technology need to be based on real social needs?

高盛全球宏观研究部的高级策略师Allison Nathan在最新的《Top of Mind》报告中提出，相对于在AI上的开发和运行上的高额投入，其经济社会效益乏善可陈。这使得围绕AI进行的资本运作可能成为一个巨大的泡沫。早前，在军备竞赛式的大模型开发热中，很业内人士就已经在不同场合表示，大模型技术发展的痛点不在于技术本身，而在于其商业化落地方面，如：如何精准针对市场需求来开发下游场景应用？如何保证长期充裕的资金和能源支持？是追求算力无限升级还是寻求有效性算力？这些问题的提出验证了在人工智能的狂热中被多少有些忽略的事实：任何一种技术的发展都离不开现实的需要和条件，而不是抽象的发展理念。
Allison Nathan, a senior strategist at Goldman Sachs Global Macro Research, pointed out in the latest "Top of Mind" report that compared to the high investment in AI development and operation, its economic and social benefits are lackluster. This makes the capital operations surrounding AI potentially a huge bubble. Earlier, during the arms race-like development boom of large models, many industry insiders had already expressed in various occasions that the pain point of large model technology development lies not in the technology itself but in its commercial implementation, such as: how to develop downstream application scenarios precisely targeting market demand? How to ensure long-term sufficient funding and energy support? Should we pursue unlimited upgrades in computing power or seek effective computing power? These questions validate the somewhat overlooked fact in the AI frenzy: the development of any technology cannot be separated from real needs and conditions, rather than abstract development concepts.

当然，“萝卜快跑”作为一种新的技术，既有它在技术上的突破，也为消费者带来了真实的利好。然而，“科技改善生活”的命题从来就不是一个简单的普遍性结论。它不是对与不对、好与不好之间的对立论证，而是不同的效益与考量之间的均衡与博弈。对于科技应用的不可逆，业界、社会、学术界和多国政府之间需要有意识地形成知识共享和最佳实践，同时关注技术安全标准和社会风险控制，通过确定人工智能安全方面最重要的开放研究问题来支持人工智能安全生态系统，以减轻各种潜在风险。它不仅关系到社会资源的投入，还关系到壁垒和鸿沟，稳定和复杂性，更关系到宏观意义的社会和谐发展。
Of course, as a new technology, "Radish Run" has both its technological breakthroughs and real benefits for consumers. However, the proposition that "technology improves life" has never been a simple universal conclusion. It is not a binary argument of right or wrong, good or bad, but a balance and game between different benefits and considerations. For the irreversibility of technological applications, the industry, society, academia, and multiple governments need to consciously form knowledge sharing and best practices, while paying attention to technical safety standards and social risk control, supporting the AI safety ecosystem by identifying the most important open research questions in AI safety to mitigate various potential risks. It not only involves the input of social resources but also concerns barriers and gaps, stability and complexity, and more importantly, the harmonious development of society in a macro sense.