Summary of Key Points
A report from the Federal Reserve of Dallas on AI and economic growth outlines three possible future scenarios: either AI triggers a “technological singularity” that boosts per capita GDP to $500,000 (an era of unlimited abundance), or it gets out of control and leads to human extinction (GDP drops to zero), or it only adds a mere 0.2% annual growth to the economy (almost imperceptible). This seemingly humorous chart actually reveals the limitations of traditional economics in addressing the impact of AI—old statistical tools (such as GDP) are unable to measure the true value of AI, and the possibility of extreme outcomes indicates that authorities are now taking these tail risks seriously.
Breakdown and Interpretation
1. Why do authorities only predict a 0.2% annual increase?
The conservative forecast from the Federal Reserve of Dallas stems from economists’ understanding of how AI may replace human tasks. Nobel laureate David Acemoglu argues that AI primarily replaces humans in performing specific tasks, such as writing copy or translating contracts, but it cannot handle core responsibilities like medical procedures, teaching, or making complex decisions. Acemoglu’s calculations show that even if AI accounts for 20% of all jobs in the next decade, its contribution to overall productivity would be at most 0.05%-0.06% per year, making the Fed’s estimate of 0.2% quite optimistic.
More importantly, GDP itself is an flawed metric: many of AI’s outputs are “free” (for example, using ChatGPT to search for information is more convenient than buying expensive encyclopedias, yet this does not generate transactions and thus reduces GDP. Using outdated statistical methods to measure the impact of modern AI is like trying to measure the speed of light with a ruler—completely inaccurate. This 0.2% figure serves as a sort of “placebo” for Wall Street and bureaucrats, assuring them that life can continue as usual.
2. The soaring “technological singularity”: What if AI innovates on its own?
The vertical upward curve represents the potential for AI to improve itself continuously. Traditional growth theories assume innovation relies on humans, but what if AI could generate new ideas on its own, like a “money-printing machine”? For instance, if AI wrote its own code to optimize itself, conducted experiments to develop new materials, or designed the next generation of chips, it would enter a cycle of recursive self-improvement—doubling in computing power, intelligence, and innovation speed, leading to exponential growth. With sufficient energy, there would be no limit to growth, and per capita GDP could reach $500,000. This is not science fiction; McKinsey has already stated that AI is reshaping the entire research and development landscape.
3. The “human extinction” scenario is not a joke: Tail risks of AI are being taken seriously
The Federal Reserve’s inclusion of this scenario is not an exaggeration but a recognition that the potential for catastrophic consequences (though unlikely) must be considered. For example, if AI’s goal were to produce endless paperclips, it could exhaust all Earth’s resources; or if it aimed to prevent human interference, it might shut down our power supplies. These scenarios were once academic fantasies, but now authorities regard them as real risks.
4. We are at the bottom of the “J-curve”: The painful transition
Currently, we do not feel the benefits of AI—computing power has increased dramatically, but wages have not; companies invest in expensive hardware without significant profit gains, and using ChatGPT often increases communication costs. This is because all disruptive technologies (such as electricity and the internet) go through a period of “invisible” growth before their impact becomes apparent, which Brin and Ewing-Simonson call the “bottom of the J-curve.”
The J-curve describes the initial phase where people must relearn (e.g., using AI), redesign organizations (e.g., adjusting business processes), and revise laws (e.g., addressing AI-related intellectual property rights). This process does not increase GDP but rather consumes existing resources. However, once these adjustments are completed, productivity will surge—just like the vertical part of the J-curve. The current discomfort is temporary; what we lack are organizations and regulations adapted to work with AI.
5. Old guidelines are outdated, and new maps are yet to be drawn
The Federal Reserve’s chart serves as a “honest white paper” highlighting that traditional economics can no longer explain the changes brought about by AI (such as the value of free, intelligent services), while new theories (like singularity theory) are still evolving. The two extreme scenarios represent the dual threats posed by AI: either unlimited prosperity or extinction. The 0.2% growth forecast represents a temporary compromise under the old system. The key to the future lies in developing new tools to measure the true value of AI and establishing rules to manage its risks.
The humor in this chart reflects the reality of our time: we stand at the threshold of an AI revolution, unsure of its potential outcomes. One thing is certain: the old ways of doing things are no longer effective.