Discussion about this post

User's avatar
Pietro Ventani's avatar

AI-driven content generation remains fundamentally tethered to the rearview mirror. Because Large Language Models are stochastically trained on historical datasets, their outputs are probabilistic averages of what has already been said. By definition, these systems are backward-looking, identifying patterns within the known rather than conceptualizing the unprecedented.

Investing and finance are a good example. In what we call the "Age of Geopolitical Dominance"—characterized by radical shifts in geopolitical alliances, energy transitions, and systemic market volatility—AI fails as a predictive tool. It cannot synthesize a future that breaks from the past; instead, it offers "hallucinations" of normalcy or recycled tropes.

For the sophisticated investor navigating a regime change where historical correlations have decoupled, AI-generated insights are not merely limited—they are potentially catastrophic, as they lack the cognitive plasticity required to anticipate a world that no longer resembles its training data. https://swerveinsights.substack.com/

No posts

Ready for more?