Introduction
In the realm of technology and artificial intelligence (AI), a recurring claim is that "all exponentials eventually become sigmoids." In other words, every exponential growth curve eventually slows down and stabilizes. This idea is often used to reassure AI skeptics: there's no need to worry about an uncontrolled explosion of AI capabilities, as everything will eventually stabilize. But how valid is this argument?
The Sigmoid Phenomenon
Simply put, a sigmoid is an "S"-shaped curve. It starts with exponential growth, then gradually slows down until it reaches a plateau. This is widely observable in various natural and technological phenomena. Take epidemics, for instance: they start slowly, the number of cases increases exponentially, then slows down as the majority of the population gets infected.
In the technological realm, consider the evolution of airspeed records. Each generation of technology (propellers, turbojets, etc.) experienced iterative improvements until reaching its fundamental limits, before being replaced by more advanced technology. The last milestone, ramjets, capped at around 3500 km/h.
Forecasting Errors of Sigmoids
However, the notion that all exponential curves will eventually morph into sigmoids has often led to erroneous forecasts. Consider the UN's demographic projections in low birthrate countries. Despite a continuous decline in births, the UN consistently forecasted a slowdown that did not occur as quickly as anticipated.
Another striking example is solar energy. For years, solar deployment forecasts underestimated the actual growth, thinking the increase would quickly stabilize. Yet, solar energy's rise has continued to surprise with its vigor, defying predictions of stagnation.
Why AI Might Not Follow a Sigmoid Pattern
AI is a complex, multidimensional field. Unlike more linear technologies, AI benefits from a multitude of factors that can extend its exponential growth. For instance, algorithm improvements, increased data availability, and constantly enhancing computing power are all elements that could delay the point at which AI would plateau.
Moreover, AI is continuously evolving through unforeseen advancements and groundbreaking discoveries. Therefore, betting on a rapid stabilization of AI could be a mistake.
The Importance of Vigilance
It's crucial not to rely on the comforting idea that AI will naturally slow down. Decision-makers must be proactive and anticipate potential AI challenges, rather than relying on historical growth models.
Conclusion
The claim that "all exponentials become sigmoids" can be misleading, especially in the context of AI. Caution and anticipation are essential to navigate the tumultuous waters of technological innovation.
Let's discuss your project in 15 minutes.