Can Safeworld convince people that gen AI robots won’t hurt them?

Robot safety simulations become materially more robust and accessible for developers this quarter.

Safeworld, a new company founded by Carnegie Mellon's Dr. Ding Zhao, former startup executive Kyle Wong, and engineer Simo Rachidi, has raised over $12 million in seed funding. The company aims to address the safety concerns associated with deploying generative AI in robots. Traditional AI algorithms are predictable, but generative AI's probabilistic nature introduces unpredictability, posing a risk to human safety. Safeworld will evaluate robotic control systems using detailed simulations populated with realistic human models, akin to how autonomous vehicle companies test their software. Their method involves creating digital twins of environments and running thousands of scenarios with simulated robots and unpredictable human behaviors, such as tripping or carrying objects. This third-party validation is seen as crucial by investors like a16z Speedrun, who believe industry safety standards need to be established now, before widespread robot deployment leads to incidents. Gritt Robotics, a company developing AI for solar panel installation robots, is already partnering with Safeworld to test its safety simulations.

The founding of Safeworld signals a significant market push towards formalizing safety standards for generative AI-powered robots. By focusing on third-party simulation and evaluation, the company addresses the growing need for trust and predictability in autonomous systems operating alongside humans. This move by Safeworld, backed by substantial seed funding, indicates that the industry is recognizing the inherent risks of deploying probabilistic AI in physical environments and is actively seeking solutions before major incidents occur.

This development directly impacts companies developing or deploying robots that utilize generative AI, as well as investors seeking to mitigate risks in this burgeoning sector. Safeworld's approach of rigorous simulation testing offers a potential path to de-risk deployments and build public confidence. The company's strategy to potentially become the first profitable entity in this niche suggests that safety validation is perceived as a necessary cost of doing business, rather than an optional add-on.