Dylan Patel describes Anthropic's unreleased Mythos model as a monumental step forward, comparing its coding ability to an L6 software engineer—a huge jump from Claude 3 Opus's L4. The capability is so advanced that Anthropic is deliberately withholding its full power, signaling a new era of model performance.

Beyond enterprise sales, the intense focus on creating AI that can code is driven by a strategic belief that this is the most direct path to Artificial General Intelligence (AGI). Leaders like Anthropic believe an AI that can recursively improve its own code will be the first to achieve superintelligence.

OpenAI and Anthropic's explicit strategy involves recursive self-improvement by creating AI that can perform ML research at a human level. They aim to scale this to millions of "AI researcher equivalents," believing this will accelerate progress far beyond competitors who rely on human talent.

AI research startup Consensus focuses its tools on automating tedious parts of science, like searching for papers, rather than trying to create a fully autonomous AI scientist. They believe the core of scientific discovery—connecting disparate ideas and human collaboration—will remain a uniquely human task.

AI generates ideas by referencing existing data, making it effective for research but poor for true innovation. Breakthroughs require synthesizing concepts from disparate fields and having a unique vision for the future—capabilities that AI lacks. It provides probable answers, not visionary ones.

Anthropic's intense focus on AI for coding wasn't just a market strategy. The core belief, held since 2021, was that creating the best coding models would accelerate their internal researchers' work, creating a powerful flywheel that improves their foundational models faster than competitors.

Anthropic's model development strategy focuses on maximizing intelligence first, accepting that initial versions may be less efficient. This approach ensures the capability frontier is always advancing, with optimization treated as a separate, subsequent step.

A key strategy for labs like Anthropic is automating AI research itself. By building models that can perform the tasks of AI researchers, they aim to create a feedback loop that dramatically accelerates the pace of innovation.

Even if AI perfects software engineering, automating AI R&D will be limited by non-coding tasks, as AI companies aren't just software engineers. Furthermore, AI assistance might only be enough to maintain the current rate of progress as 'low-hanging fruit' disappears, rather than accelerate it.