About a month ago, there was an invitation to appear on a particular podcast called THE NETWORK STATE. The podcast is related to my interests in complex adaptive systems, investing, AI, and what can be modeled. This was my seventh or eighth podcast appearance, and it has helped me refine how I communicate my ideas verbally. I was not completely satisfied with some of my responses regarding why I strongly support e/acc. This experience has motivated me to provide a better explanation of my thought framework around topics such as complex adaptive systems, threats, AI, and the limitations of modeling.
The Significance of Models
Models play a crucial role in predicting future outcomes. They provide a sense of certainty in an uncertain world, which is something that humans generally prefer. Models offer a way to look into the future and make it more predictable, thereby reducing uncertainty. However, there are certain aspects that cannot be accurately modeled, and this can be frustrating. Despite this challenge, we often attempt to model these complex phenomena.
I have a great appreciation for models, particularly in the context of investing. For instance, I value financial models like the DCF model used for analyzing companies such as Walmart or insurance companies. These types of companies have a track record that allows for reliable predictive modeling based on historical data. On the other hand, I am not as interested in seeing models for companies like Cloudflare, Nvidia, or Ethereum. These types of technological investments involve significant uncertainty and unpredictability, making them challenging to model accurately.
Investing in growth tech, like cryptocurrency, requires an understanding of S-curves and power-law dynamics, which are difficult to capture in a traditional financial model. Predicting human adoption trends and new behaviors is complex and cannot be fully modeled.
It's essential to acknowledge the limitations of modeling when dealing with complex adaptive systems. While models can provide valuable insights, they often involve educated guesses about the future. These extrapolations are based on known inputs but may lead to unexpected outcomes due to the complexity of the system.
Challenges with AI Doomer Rationale
The concept of AI doomerism is based on a series of assumptions and extrapolations that have a very low accuracy rate. When these assumptions compound over multiple levels, the likelihood of accurate predictions diminishes significantly. This approach may overestimate the potential risks associated with AI, manifesting in doomsday scenarios that are improbable.
It is crucial to assess the actual threats posed by AI realistically. Governments and bureaucratic entities have a long history of misuse of power and incompetence, which may pose a more significant risk than the doomsday scenarios projected by AI skeptics.
Placing excessive control over AI in the hands of government institutions may not guarantee safety. In contrast, innovative engineers focused on advancing AI technology for the benefit of humanity may present a more promising approach. The likelihood of AI being misused by power-driven individuals in government positions is a tangible concern that should not be overlooked.
In conclusion, embracing a realistic and balanced perspective on the potential risks and benefits of AI is essential. Avoiding extreme doomsday scenarios and acknowledging the potential dangers of centralized control over AI can lead to a more informed debate on the subject. By considering the patterns in historical abuses of power, we can better understand the genuine threats posed by AI and make informed decisions moving forward.
