In order to work, our algorithms must first be trained on a curated set of structured data, that is to say, data annotated by hand. The annotation phase assigns one or more labels to the elements of a dataset to make it possible to train the models used in our applications. It is therefore a service that we systematically use during the configuration phase of our new customers.
The need for annotation services is growing exponentially in line with the success of artificial intelligence, sometimes leading to abuses. The working conditions of these “click workers” have also recently gained visibility through academic publications and general public reports.
The relevance and performance of a solution strongly depends on the quality of the data and the quality of the labels. Deepomatic therefore paid particular attention to the choice of our annotation partner.
Taking a closer look at this market, we discovered that it rarely benefits the local economies in which it is established. This is why, as soon as it was possible for us, we supported the creation of local businesses.
This choice allows us to maintain a privileged relationship with one of our main stakeholders. Today we are proud to partner with Shakti Outsourcing.
In addition to the search for quality, knowing our employees allows us to fully exercise our duty of vigilance. In other words, to ensure that our requirements in terms of working conditions and human rights are fully met.
Today we want to contribute to a more global reflection on the issues specific to the development of this new form of subcontracting in a globalized world. This is why we have chosen to open our doors to a large French laboratory working on these questions.