Introducing Inter-1, multimodal model detecting social signals from video, audio & text

About this article

Hi - Filip from Interhuman AI here 👋 We just release Inter-1, a model we've been building for the past year. I wanted to share some of what we ran into building it because I think the problem space is more interesting than most people realize. The short version of why we built this If you ask GPT or Gemini to watch a video of someone talking and tell you what's going on, they'll mostly summarize what the person said. They'll miss that the person broke eye contact right before answering, or pa...