Summary of Key Points
The article uses the metaphor of "firewood and welding" to illustrate that AI technology is like welding—highly energetic and capable of solving complex problems—but traditional economic indicators (such as GDP and productivity) struggle to reflect its value. Traditional technologies, on the other hand, may have a lower energy output, but they can provide warmth to the surrounding environment (corresponding to growth reflected in traditional economic data). By revisiting the "Solow Paradox" (the phenomenon where investments in the computer era did not lead to increased productivity), the article points out that this paradox is even more pronounced in the AI era. Free digital services (such as WeChat, TikTok, and ChatGPT) offer great convenience, yet their value remains invisible within the traditional statistical system. AI not only challenges GDP calculations but may also subvert the price-based foundation of the economy, leading to deflation and a new understanding of the AI bubble: despite significant investment, their benefits are difficult to quantify using traditional data, and the bubble could burst, yet the changes they bring are real.
Detailed Analysis
1. Firewood and Welding: The Unique Impact of AI on the Economy
The difference between firewood and welding highlights the contrast between traditional technologies and AI:
- Firewood: The flame is spread out over a large surface, radiating heat through infrared radiation, which we can feel as warmth (temperatures range from 600-800°C). This is similar to traditional technologies that can boost the overall economy, leading to visible increases in indicators like GDP and employment rates, and people can directly experience improved living standards.
- Welding: The temperature is extremely high (over 3000°C), capable of cutting steel, but the energy is released in the form of intense light and ultraviolet radiation, with little increase in air temperature, making it less noticeable to humans. AI is like welding; it can perform complex tasks (such as writing code or diagnosing diseases), but the value it creates is not in the form of traditional indicators like GDP or prices, so traditional economic data cannot capture its impact.
The core message of this metaphor is that the value of AI is fundamentally different from that of traditional technologies and cannot be measured by the same metrics.
2. The Re-emergence of the Solow Paradox: The Value of AI is Invisible in Data
In 1987, Paul Solow observed that despite widespread computer adoption, productivity did not increase, leading to the "Solow Paradox." This paradox is even more evident in the AI era:
- We use free services like WeChat for communication, TikTok for entertainment, and ChatGPT for information, which save time and enhance our experience. However, traditional GDP statistics only account for "things that cost money." Free services, being non-traditional transactions, are not included in GDP calculations.
- For example, a Stanford professor noted that switching from buying physical newspapers (which contributes to GDP) to consuming free online news can result in a decrease in GDP, even though user experience improves. This illustrates how the value of AI cannot be accurately reflected in economic data.
3. Why Does Traditional GDP Statistics Not Capture the Value of AI? Free Services Are the Key
GDP is essentially the total value of market transactions, but AI has created a large number of free services whose value cannot be measured:
- WeChat allows us to contact people for free, saving on phone calls and writing time, but these savings are not included in GDP.
- ChatGPT helps us find information quickly, saving time compared to searching through books, but this time savings cannot be valued.
- TikTok provides free entertainment, replacing paid services like movies and magazines, yet their value is not reflected in GDP.
In other words, traditional GDP only recognizes transactions that involve money, while AI makes many things available for free, leading to a gap in the statistical system.
4. Could AI Lead to Deflation?
The core of the modern economy is the "price system"—GDP is calculated based on prices, and businesses calculate profits based on prices. However, AI is bypassing this system:
- Sam Altman, CEO of OpenAI, suggests that the AI economy is inherently deflationary, as AI makes products cheaper or even free. For example, AI-generated content replaces paid writers, reducing prices.
- Canadian entrepreneur Jeff Bush argues that technological advancements lower costs and drive down prices. AI can automate many tasks, eliminating intermediate steps (such as online shopping, avoiding rental costs for physical stores), leading to deflation.
- The article emphasizes that convenience often comes from bypassing these intermediate steps, which can result in lower prices and deflation.
The issue with deflation is that continuous price declines can reduce corporate profits, potentially affecting investment and employment, posing a challenge to the economic foundation.
5. A New Perspective on the AI Bubble
The term "AI bubble" arises because significant investment in AI has not resulted in corresponding increases in traditional economic indicators. However, the article offers a different view:
- From a traditional data perspective, billions have been invested in AI research and development, but GDP has not increased, and businesses have not seen substantial profits, resembling a bubble.
- From a utility perspective, the benefits of AI (free services, time savings, improved experiences) are real, but they are not measurable in terms of prices.
- The article concludes that the AI bubble may burst (e.g., companies investing in AI may see their stock prices drop), but the changes AI brings to the economy and society are real, just as welding cuts through steel, even if it does not directly generate visible heat.
Conclusion
AI is not like traditional technologies; it is more akin to welding—highly impactful but invisible to traditional metrics. This is not just a problem with the statistical system but also a shift in the logic of the economy. Prices are no longer the sole measure of value, and intangible benefits such as free services and time savings are becoming increasingly important. In the future, we may need new indicators to assess the impact of AI. Otherwise, we will face a situation where life improves, yet economic data fails to reflect these changes. The potential for deflation and the associated risks of an AI bubble require us to reevaluate the rules of the economy.