Machine learning models can be trained on large datasets of resumes and job descriptions, which can help to reduce the potential for explicit bias in the screening process. However, it’s important to note that the data used to train the model may itself contain implicit biases. If the data used to train the model is biased, the model will also be biased in its predictions. Therefore, it is important to carefully examine and clean the training data to minimize any existing biases before training the model.

It is also important to note that bias can be introduced in other ways than the data, like in the feature engineering process, or in the algorithm’s architecture. Therefore, it’s crucial to evaluate the model’s performance with diverse groups of candidates and monitor performance metrics to detect any potential biases.

Moreover, the use of machine learning models alone is not enough to remove explicit bias in the recruitment process. A thorough human review and monitoring should be done in parallel to detect and correct any potential issues.