Google open-sourced MT5, a model trained on over 100 languages that the company says achieves state-of-the-art benchmark results.
TechSee raises $30 million to streamline field service work with AR and computer vision
TechSee, a startup employing AI and computer vision to streamline field service work, raised $30 million to expand its reach.
MIT CSAIL’s Roboat II is an autonomous platform large enough to carry human passengers
Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory have developed an autonomous boat large enough to carry human passengers.
Ketos raises $18 million to monitor drinking water quality with AI
Ketos, a startup leveraging AI to identify water quality problems, has raised $18 million — a mix of equity and debt.
Automatic signature verification software threatens to disenfranchise U.S. voters
An increasing number of U.S. counties are using signature verification software to check ballots. But the software could exhibit bias.
Researchers design virtual environment to spur development of helpful home robots
Researchers developed an environment to test embodied AI agents on their ability to understand goals and help accomplish them.
AI Weekly: Constructive ways to take power back from Big Tech
Here is some advice on how everybody can take power back from Big Tech companies like Facebook and Google.
AI researchers urge tech go beyond scale to address systemic social issues
Scale isn’t just a business plan but an approach that makes Big Tech and startups less effective at addressing systemic social issues.
Researchers find high error rates in commercial speech recognition systems
In a study, researchers found the error rates of several commercial speech recognition models to be higher than previously assumed.
Microsoft and MITRE release framework to help fend off adversarial AI attacks
Microsoft, MITRE, and 11 other organizations developed a framework to help fend off adversarial machine learning attacks.