Exclusive: Startup teaches AI to read emotions
Hume AI, a small New York startup, has raised $12.7 million on the premise that it's not enough for AI systems to understand the world's information — they also need to understand human reactions.
Why it matters: AI startups are all the rage, but most are focused on mimicking humans' critical thinking rather than detecting their emotions.
Between the lines: CEO Alan Cowen said the goal of his technology is to help automated systems recognize when humans are pleased or frustrated.
- "It’s not really feeling things in your place," Cowen tells Axios. "It’s about understanding what you are going through."
The intrigue: Combining powerful AI technology with an ability to detect human emotions could help AI-based systems work more effectively. However, it also raises alarms that the systems could eventually be used to play on those emotions.
- Cowen says the company has put in place safeguards and requires its customers agree to a set of principles over how the technology will be used.
- Specifically, Hume mandates its technology can't be used for surveillance or manipulation, and that people are never tricked into thinking they are conversing with another person.
- "AI should be optimized not to buy things or click things, but to make us happier," says Cowen, who was a computational scientist specializing in emotions before starting Hume.
Details: The cash infusion will allow the 13-person company to staff up and comes as it seeks to expand a private beta of its technology, already in use by a few hundred companies in areas including call center analytics, health technology and AI research, Cowen said.
- The funding round was led by Union Square Ventures, with participation from Northwell Holdings, Comcast Ventures, LG Technology Ventures, Wisdom Ventures and Pinterest c0-founder Evan Sharp.