Super.AI offers two options when it comes to choosing who labels your data:
Which option you choose depends on your project’s requirements. Below, we examine the benefits of using super.AI, as well as some of the reasons you might opt to use your own labelers. You’ll also get an insight into the labeling interface that people working on your data use.
The super.AI crowd provides immediate access to fast, reliable results. Our labelers undergo qualification tests and continuous monitoring to ensure quality (you can learn about this in detail in our quality assurance blog post). All of this means we can provide a quality, cost, and speed guarantee tailored to your project across all the data we label. We cannot provide this guarantee when you use your own labelers; we can only measure the quality of their output.
An integral part of the super.AI system is that it intelligently routes tasks to multiple labeling sources to achieve the best possible results. In addition to our human labelers, we employ third-party sources and train AI to label your data where possible. This automated process leads to quicker, cheaper outputs. You can learn more about these processes in our AI compiler blog post. When you opt to use your own labelers, the super.AI meta AI and third-party solutions are excluded as labeling sources.
There are two primary reasons why you might choose to use your own labelers on a project. Let’s take a look at each one in detail.
Super.AI follows stringent data privacy regulations. You can find a full rundown of how we handle your data in our documentation: data privacy and security. However, we know that often data can be so sensitive that it cannot be handled by third parties at all. If your data has to stay in house, the option to use your own labelers allows for this. When you use your own labelers on the super.AI platform, your data will not be viewed by anyone outside of whom you invite to your project.
At any time, you can download information on any of your data points, which will include who uploaded the input data, who processed the data point, who viewed the data, and who reviewed the data point’s output.
The super.AI crowd is trained to handle a huge variety of tasks and our labelers process data with guaranteed results. But the truth is that some projects demand specialized knowledge that can’t be outsourced to a third-party crowd. When you choose to use your own labelers on a project, you can handpick the labelers you need to provide processing expertise.
When you choose to label yourself, the project owner and everyone they invite to the project has access to the super.AI labeling user interface (UI). This is where users are able to process data points. The appearance and the tooling of the UI is customized to the project type, but there are several features that remain consistent.
There is the option to provide feedback on difficult or ambiguous data points. Labelers also have the option to skip tasks if they are not possible to complete. When this happens, the labeler can provide a reason why the task is impossible, so that the project owner can improve the input or task design.