Superhuman 的 AI 草稿 60% 無需修改 —— 但我們仍無法完全信任它
電子郵件客戶端 superhuman 推出了一項新的自動草稿功能，使用了來自 openai 和 anthropic 的前沿模型。該公司聲稱 60% 的發送草稿不需要編輯，儘管準確性仍然不穩定。
email client superhuman released a new version of its auto-draft feature, utilizing a mix of frontier models from openai and anthropic to generate replies that mimic user tone. beta testing data indicates that 40% of drafts were sent within 24 hours, and 60% of those were sent without any edits. the system learns from user behavior over time to correct mistakes, such as refusing unreasonable meeting times, but still requires oversight. this shift moves beyond older gpt-3.5 implementations to offer higher intelligence, signaling a trend where ai agents handle increasing amounts of routine communication.