Anthropic preps model that better reads users' intentions

Anthropic is preparing a new model that better discerns users' underlying goals than its current commercial models, even with little explicit information. A tester said it identifies root causes of problems rather than single fixes and generates its own approaches instead of generic advice.

• Model infers user intent even with minimal input
• It spots a whole class of bugs and asks whether to fix all
• It devises its own algorithms instead of generic tips
• OpenAI, Anthropic and Google shipped over a dozen models in a month