Nicholas Marwell, Member of Technical Staff of Anthropic, shares predictions for Generative AI models as well as guidance on safely training and scaling them for the near future.

Topics Include:

• Where are Generative AI Models heading
• About Anthropic
• Claude Opus, Claude 3 Sonnet & Claude 3 Haiku
• Scaling laws impact on models
• Training models for text, images and multimedia
• AI Safety, high stakes and stewarding through potential dangers
• Constitutional AI and interpretability research
• Benefits of Features – combinations of neurons
• Final thoughts – high growth and pace for coming months and years