Independent Essay on OpenAI Platform Outlines Two AI Futures: Depth vs. Width

Hemanth Asirvatham and Elliott Mokski authored an independent essay titled The eternal complement, hosted on a new OpenAI platform for exploring AGI futures. The essay explicitly notes it reflects the authors' views, not OpenAI's. It explores two divergent paths for AI-driven progress: Civilization of Depth, where intelligence overcomes physical constraints through simulation and reasoning, and Civilization of Width, where progress is limited by the slow pace of physical execution and institutional bureaucracy.

The Rising Cost of Progress

The authors cite research by Nick Bloom and others to highlight the increasing difficulty of scientific advancement. Key metrics include:

• Maintaining Moore's Law now requires 18x more researchers than in the early 1970s.
• Since the 1930s, effective research effort increased 23x, yet measured productivity fell to 1/41st.
• Current chip fab costs are 5x higher than 30 years ago.

Depth vs. Width

The essay posits that Frontier intelligence and Capacity are complements. In a Civilization of Depth, superintelligence uses simulation to narrow uncertainty, allowing for efficient physical experiments (e.g., predicting the Higgs boson). This results in deep wisdom with a small physical footprint.

Conversely, a Civilization of Width faces bottlenecks in fields like biology, where in silico models cannot fully replace empirical testing. Here, the true power of AI may not be genius, but the ability to act as a bureaucrat, coordinating vast mechanical and institutional systems. Physical expansion, such as building a Dyson sphere, remains constrained by the speed of light and material logistics, potentially stretching progress over millennia.

The Role of Human Curiosity

Despite AI's potential superiority in insight and resource coordination, the authors argue human curiosity remains vital. Humans may retain a comparative advantage in Frontier intelligence and creative diversity, while AI handles the "boring" work of execution. The future depends on whether progress requires more material resources (Width) or achieves more with less (Depth).