AI Intelligence Growth Slows: Hedge Fund Data Shows Plateauing Effect

Data from an AI hedge fund indicates that AI intelligence gains are plateauing, challenging the narrative of a rapid AI revolution. While large language models (LLMs) reached a useful level of intelligence in 2024-early 2025, subsequent gains have slowed. Elo ratings, used to measure AI intelligence, have decreased from approximately 44 points per quarter in early 2025 to around 13 points in February 2026. Furthermore, the cost of training frontier AI models is projected to increase significantly, reaching $1.2 billion by early 2027 for a marginal improvement of 6 Elo points. This suggests diminishing returns on investment in raw AI intelligence. The top AI models are also converging in performance, with only a small difference in Elo ratings. While AI agent task completion time has improved exponentially, this is not necessarily indicative of increased intelligence. These trends suggest that the market's expectation of a rapid AI-driven transformation may be unrealistic, and that integration and redesign efforts will take longer than anticipated. The focus should shift towards practical applications and efficient deployment, rather than solely pursuing marginal intelligence improvements.

Transparency is critical in AI development and deployment. As AI systems become more integrated into our lives, it's essential to understand how they work and what data they use. This analysis is based on publicly available information and aims to provide a clear and objective assessment of the current state of AI. It is important to note that AI is a rapidly evolving field, and future developments may significantly alter the landscape.

This deep analysis is intended to comply with EU Article 50, ensuring transparency and providing a clear understanding of the AI's capabilities and limitations.