The AI development cycle of experimentation and bottleneck-solving is already a form of recursive self-improvement. Kyle Corbitt argues this loop is currently constrained by human intelligence. Once AIs become better at directing this process, progress will accelerate rapidly.

Progress in AI isn't a smooth, continuous line. Just as Moore's Law required discrete inventions, AI scaling relies on paradigm shifts. The current Transformer+RL approach may hit diminishing returns, and an AI trained within this paradigm is unlikely to discover the next fundamental breakthrough required to maintain progress.

AI is not a 'set and forget' solution. An agent's effectiveness directly correlates with the amount of time humans invest in training, iteration, and providing fresh context. Performance will ebb and flow with human oversight, with the best results coming from consistent, hands-on management.

Current LLM agents are effective at executing and optimizing experiments within a defined research track, like hyperparameter tuning. However, they lack the crucial scientific skill of 'lateral thinking'—recognizing when a research path is a dead end and strategically pivoting to a fundamentally new approach.

As AI agents eliminate the time and skill needed for technical execution, the primary constraint on output is no longer the ability to build, but the quality of ideas. Human value shifts entirely from execution to creative ideation, making it the key driver of progress.

AI agents can flawlessly execute predefined tasks (SOPs). However, they still require significant human management to ensure high-quality output, apply taste, and surface meaningful signals from the data they generate. This creates a new layer of human work, rather than a complete replacement.

The future of company operations involves creating cascading AI loops for various functions. These loops optimize tasks to a local maximum, but then plateau. Human intuition and out-of-distribution thinking are critical to initiate the jump to the next level of innovation, or the 'next hill'.

AI can generate hundreds of statistically novel ideas in seconds, but they lack context and feasibility. The bottleneck isn't a lack of ideas, but a lack of *good* ideas. Humans excel at filtering this volume through the lens of experience and strategic value, steering raw output toward a genuinely useful solution.

Building an AI agent is the starting point, not the finish line. The real, ongoing work lies in optimizing its performance and training it on new information. This creates an essential new human-in-the-loop role focused on continuous improvement.