The design of AI interfaces is evolving towards natural language interactions, allowing users to collaborate with AI as if it were a co-worker rather than just a tool.

The future of AI will likely involve a shift from knowledge labor to users orchestrating AI systems to enhance their effectiveness.

Many mundane tasks in information processing are perfect for automation using large language models, yet there are few applications targeting this area.

Achieving human-level AI will not be possible by merely scaling up existing large language models; a fundamental shift in approach is necessary to reach that goal.

Computer use represents a significant advancement for AI, allowing models to adapt to tools rather than requiring developers to create custom environments for each task.

The AI field is approaching diminishing returns with large language models, as the amount of natural text data available for training is becoming limited, which may hinder future advancements.

At Anthropic, while our models are intelligent, we must communicate how they can be useful and connect with users, not just highlight their intelligence. We need to design AI tools like Claude to be genuinely useful for people discovering large language models for the first time, providing them with concrete, practical applications in their daily lives.

AI has transformed our business, enabling us to offer conversation practice through large language models, which was previously impossible without human interaction.

The last two years have been dominated by large AI models, which have proven to be popular and effective, but we are now entering an era focused on complex workflows around these models.