Recently, pessimistic predictions related to AI have become a trending topic online. Indeed, disruptive impacts are unavoidable. However, based on my long-term observation of the market, the pace of these changes is much slower than everyone imagines.

First, do not underestimate the inertia of existing systems. The peak oil theory of 2007, the collapse of the dollar system in 2008, the end-of-AMD and NVIDIA predictions in 2014—each time, people declared the end of the current system, but reality has always betrayed those predictions. Institutions with deep inertia are far more resilient than outsiders imagine.

The example of real estate agents is typical. For the past 20 years, there have been calls for the "end of real estate brokerage." Even with platforms like Zillow, Redfin, and Opendoor emerging, the market’s inertia and regulations have protected this profession, allowing it to survive much longer than expected. I actually bought a house a few months ago, and hiring an agent was mandatory; he earned about $50k. The actual work involved just paperwork and coordination, which could be done in about 10 hours at most. While efficiency will inevitably improve, it takes a long time.

Next, the mediocrity of the current software industry is more than I imagined. From my experience investing hundreds of thousands of dollars in Salesforce, these products are truly terrible. They are riddled with bugs and even mobile support is unsatisfactory. Stripe and Linear are favored simply because they are not as difficult to use as their competitors.

A deep truth is hidden here. Even if a software singularity occurs, the demand for human software labor is nearly infinite. The most effort is required to perfect the last 5 percent of quality. Almost all existing software has at least 100 times more room for improvement in complexity and functionality before reaching saturation. As a programmer in 2020, my productivity was equivalent to hundreds of programmers in 1970, yet there is still significant room for optimization. This is the "Jevons paradox"—improving efficiency often leads to explosive growth in overall demand.

Labor transfer will certainly happen. However, the U.S. has infinite potential for re-industrialization. The country has almost completely lost basic manufacturing capabilities for batteries, motors, and semiconductors, and the entire power industry depends on overseas sources. Do you know that China produces 90% of the world's synthetic ammonia? If supply is cut off, fertilizer cannot be made, leading to famine.

Physical infrastructure has no "singularity." It is constrained by frictional forces. Rebuilding bridges and roads, constructing seawater desalination plants, and maintaining them long-term—all these areas have infinite demand. A senior product manager who lost an $180k annual salary at Salesforce might find new opportunities in these large-scale projects.

The speed of change varies across economic sectors. Transformations in nearly all fields will be slower than the pessimistic predictions suggest. By accurately understanding the mediocrity and limitations of AI, and not underestimating the inertia of institutions and the complexity of human society, sufficient time for adaptation can be created.

What matters most is ensuring material prosperity in people's lives. If AI compresses profit margins to zero, consumer goods will become extremely cheap. Governments also have room to implement large-scale stimulus measures if needed. As long as we respond sensitively, ultimately, safety can be secured.