罗伯特·艾伦·费尔德曼（Robert Alan Feldman）是摩根士丹利MUFG证券公司的资深顾问。
Robert Alan Feldman is a senior adviser at Morgan Stanley MUFG Securities.
人工智能（AI）引发了截然不同的争论：乐观主义者认为它将开启一个繁荣的新时代；悲观主义者则担心它会摧毁大量工作岗位并破坏社会稳定。然而，双方的讨论都偏离了问题的核心。
Artificial intelligence has split the debate in two. Optimists say it will usher in a new era of prosperity. Pessimists say it will destroy jobs and destabilize society.
真正的问题不在于人工智能本身能做什么，而在于我们的经济体系会如何应对这一技术变革。
Both sides are asking the wrong question. The real issue is not what AI do, but how our economies respond.
目前关于人工智能与劳动力关系的多数讨论都集中在哪些任务可以被自动化上，这种视角过于狭隘。实际上，自动化会改变就业结构，进而影响人们的收入；收入的变动又会进一步改变人们的消费行为，进而迫使企业调整生产和招聘策略。这些反馈机制最终将决定人工智能是会推动社会进步，还是会带来负面影响。
Most commentary on AI and labor fixates on which tasks can be automated. This approach is too narrow. Automation changes jobs. Job changes alter incomes. Altered incomes reshape spending. Reshaped spending forces firms to adjust output and hiring. Those feedback loops will determine whether AI lifts society or fractures it.
那么，影响社会发展的关键因素是什么呢？一个针对日本经济状况制定的宏观经济模型表明：同样的人工智能技术进步，可能会带来两种截然不同的未来结果：在一种情况下，人工智能会带来适度的通货膨胀、就业率的上升以及实际工资的提高；而在另一种情况下，它则可能导致通货紧缩、就业率急剧下降以及实际工资的下跌。
What are the key pressure points? A macroeconomic model calibrated to Japan shows that the very same AI-driven productivity gain can produce two radically different futures. In one, AI delivers modest inflation, rising employment, and higher real wages. In the other, it triggers deflation, collapsing employment, and falling real wages.
虽然技术本身没有改变，但最终的结果却大相径庭。有四种力量将决定我们走向哪种未来：首先，市场竞争环境至关重要。如果占主导地位的企业面临较少的竞争压力，它们更有可能将人工智能带来的生产率提升转化为企业利润，而非用于扩大生产规模，从而导致新工作岗位的减少。而在竞争更为激烈的市场中，企业会受到外部压力而不得不扩大生产，从而创造更多就业机会。
The technology change is the same. The outcome is not. Four forces will decide which path we take. First, competition. If dominant firms face little competitive pressure, they are more likely to absorb AI's productivity gains into profits than to expand output. That means fewer new jobs. In more competitive markets, firms are pushed to grow, and growth creates work.
例如，即使人工智能降低了食品行业的成本，但由于竞争减弱，这些成本节省最终仍会流入企业利润，而不会直接体现在食品价格的下降上。
For example, even if AI lowers costs in the grocery business, lower competition means that AI gains will end up in corporate profits, not lower food prices.
第二，消费。生产率提升只有转化为收入增加，才能带动需求。如果家庭和企业将人工智能带来的收益用于扩大产能或开发新产品，需求就会上升，就业就能保持稳定。如果它们将这些收益闲置起来，需求就会减弱，劳动力市场就会萎缩。日本企业往往坐拥巨额现金。如果它们提高工资和/或增加企业投资，工人就会有更多钱可花，人工智能带来的收益也能扩散，甚至惠及人工智能使用程度较低的领域。
Second, spending. Productivity gains help demand only if they raise income. If households and firms spend or invest AI gains in more capacity or new products, demand rises and employment holds up. If they sit on them, demand weakens and the labor market shrinks. Japanese companies often sit on large piles of cash. If they were to pay higher wages and/or raise business investment, workers would have more to spend and the benefits from AI would spread, even to sectors where AI use is low.
日本HighLander开发的仿人机器人“N”亮相于2026年9月9日在日本东京举行的生成式人工智能加速器挑战赛（GENIAC）活动，该活动由日本经济产业省（METI）主办。
"N", a humanoid robot developed in Japan by HighLander, is displayed at the Generative AI Accelerator Challenge (GENIAC) event, hosted by Japan's Ministry of Economy, Trade and Industry (METI) in Tokyo, Japan, September 9, 2026.
第三，适应能力。劳动者不经过再培训，就无法转向新的岗位。快速开展再培训可以缓冲冲击，调整迟缓则会使转型沦为社会错位。而最有效的再培训项目往往并非由政府提供，而是来自愿意投资于自身员工的企业。新英格兰早期的纺织厂甚至也开展了广泛的教育培训项目，以确保工人能够在生产车间发挥更多创新精神。
Third, adaptability. Workers cannot move into new jobs without reskilling. Rapid reskilling softens disruption. Slow adjustment turns transition into dislocation. And the most effective reskilling programs often come not from governments, but from firms willing to invest in their own people. Even the early spinning mills in New England ran broad-based education programs so that their workers would be more innovative on the factory floor.
第四，分配。如果劳动者公平分享生产率提升带来的收益，他们就更有可能接受并推动技术变革。否则，他们更有可能成为现代卢德分子——19世纪英国纺织工人运动中的成员，该运动曾抗议使用某些自动化机械。税收政策同样重要。实际到手收入比表面工资涨幅更重要。
Fourth, distribution. If workers get their fair share of productivity gains, they are far more likely to embrace and accelerate technological change. If not, they are more likely to become modern-day Luddites -- members of a 19th-century English textile workers' movement that protested the use of certain automated machinery. Tax policy matters, too. Take-home pay matters more than headline wage gains.
如果这四项条件具备，人工智能无需太多政策刺激，就能提振增长、提高工资并支撑就业。但如果市场过度集中、需求疲弱、培训进展太慢，且收入增长惠及面过窄，人工智能可能加剧通缩和经济收缩。在这种情况下，政府不能袖手旁观，而必须出手干预——而且所需干预规模扩张的速度，可能快于生产率提升本身。
If these four conditions are in place, AI can raise growth, lift wages and support employment without much policy stimulus. But if markets are too concentrated, demand too weak, training too slow and income gains too narrowly shared, AI can become deflationary and contractionary. In that world, government cannot stand back. It must step in -- and the scale of intervention required may rise faster than the productivity gains themselves.
这引出一个关键问题。仅看经济增长，无法判断人工智能是否有助于社会。国内生产总值可能增长，但工资停滞、就业走弱，焦虑与不安全感不断蔓延。真正重要的指标是工资、价格和就业。这些指标将告诉我们，人工智能带来的是广泛共享的繁荣，还是仅仅在加深既有的社会裂痕。
That leads to a crucial point. Economic growth alone will not tell us whether AI is helping society. Gross domestic product can grow while wages stagnate, employment weakens and insecurity spreads. The indicators that matter are wages, prices and jobs. Those will tell us whether AI is generating broadly shared prosperity or simply deepening existing fault lines.
人工智能更有利于灵活、竞争且包容的经济体，同时也会暴露僵化、集中且自满的经济体固有的问题。
AI rewards economies that are flexible, competitive and inclusive. It exposes those that are rigid, concentrated, and complacent.
日本具备从人工智能发展中获益的良好条件。近几十年来，日本在许多行业中的竞争力有所增强，例如高附加值农业、能源、金融和分销领域。公司治理改革正促使日本企业减少现金持有量，进而推动收入和经济增长。随着社会对换工作的偏见减弱，以及年轻人从跳槽中获益，职场流动性大幅提升。在收入分配方面，劳动力短缺正推动年轻人和兼职人员的工资以高于平均水平的速度上涨，表明收入分配可能趋于更加平等。
Japan is well placed to benefit from AI. In recent decades, Japan has become more competitive in many industries: e.g., in high-value agriculture, energy, finance and distribution. Corporate governance reforms are pressuring Japanese companies to reduce cash holdings, and thus to spur income and growth. Job mobility has surged, as the social stigma associated with job changes has fallen, and as young people benefit from job hopping. On distribution, the labor shortage is spurring above-average wage increases for the young and for part-timers, indicating that income distribution may be moving toward more equality.
人工智能带来的挑战主要不在技术层面，而在政治、制度和个人层面。能够适应变化的社会将繁荣发展；无法适应的社会则可能把实际上由自身造成的失败归咎于他人。
The challenge of AI is not mainly technological; it is political, institutional and personal. The societies that adapt will prosper. The ones that do not may blame others for failures that are actually their own.