Anthropic CPO Mike Krieger on Claude Opus 4, AI Adoption, and Future of Work

Claude Opus 4 and Sonnet 4 Adoption

• 🚀 Claude Opus 4 and Sonnet 4 models have seen incredible adoption in their first two weeks, with over 80% of customers already moving to the new versions.
• 📈 On the consumer-facing product cloud.ai, new sign-ups for pro and max accounts have doubled or tripled, indicating strong user excitement.
• 💡 The models are designed to meet users where they are, with Opus being the most powerful for complex problems and Sonnet as a versatile everyday model.

Differentiating AI Models and Partnerships

• 🎯 Anthropic aims to customize AI for specific applications, recommending Opus for intensive tasks like pharmaceutical research and Sonnet for continuous use by engineers.
• 🧩 Successful customers often utilize multiple models from Anthropic's family, integrating AI into various business functions.
• 🤝 Enterprises are seeking partnerships with AI companies, valuing trust, teams, and professional services alongside advanced models.

Training and Data Ingestion

• 🧠 Advancements in AI are occurring on both pre-training (what the model knows) and post-training (how it applies knowledge) stages.
• 🛠️ Significant progress has been made in post-training, focusing on making models proficient in specific tasks like engineering, writing, and solving complex problems over extended periods.
• ⚠️ The use of data for training is being addressed, with Anthropic emphasizing adherence to standards like robots.txt and providing citations and links in its products.

Impact of AI on Labor Markets

• 💬 The conversation around AI's impact on labor markets is crucial, with a need for dialogue on reskilling and policy solutions.
• 📈 Current usage of Claude primarily involves augmentation of tasks rather than full job replacement.
• ⏳ While AI may eventually lead to more replacement, the current years are seen as an opportunity to make AI-assisted work highly efficient and prepare for future shifts.