Making data more human

Quality is at the very center of what we do. Of course, when it comes to AI – with the
goal to help machines behave not like robots, but like people – the term “quality” is
more synonymous with “humanity” than perfection. That’s why our quality measures
combine the power of technology with natural human behavior in a real-world setting,
while ensuring the highest standards of accuracy and authenticity.
Here are some of the best practices and processes we use to deliver the highest quality
data.

Step 1

Gold standard data sets

We integrate real time audits into project workflows as a way of controlling quality during the course of a job. This involves the use of Gold Standard Tasks, for which there is a previously known correct answer agreed by domain experts and based on “ground truth” for a given project. Gold tasks are strategically (and silently) assigned along with regular tasks, with the purpose of automatically assessing contributors as they complete long sequences of tasks.

Learn More

Step 2

Gold standard data sets

We integrate real time audits into project workflows as a way of controlling quality during the course of a job. This involves the use of Gold Standard Tasks, for which there is a previously known correct answer agreed by domain experts and based on “ground truth” for a given project. Gold tasks are strategically (and silently) assigned along with regular tasks, with the purpose of automatically assessing contributors as they complete long sequences of tasks.

Learn More