Ask these 3 questions to improve Machine Learning Ethics
It's not easy to navigate Machine Learning (ML) ethics as a Product Manager. But the below three questions can help you be more mindful of biases and promote privacy & data protection throughout your ML lifecycle.
Who is building the algorithms?
Is the data set comprehensive?
Is your data biased?
Encourage your team to discuss these so you can continue to solve customer problems fairly and sustainably.
Who is building your algorithms?
Your system is the direct reflection of who built it. A diverse team will be reflected in the end product. The best way to prevent biases is by ensuring that the designers and engineers working with the ML system incorporate cultural and inclusive dimensions of diversity from the start.
Is the dataset comprehensive?
An ML model is only as good as the data used to train it. Evaluate if your data is comprehensive and representative of different variables. For example, is your data relevant to the problem you are trying to solve? Suppose you are designing a machine learning algorithm for a running app. In that case, you will have no need even for the best of datasets that consist of celebrity photos. Furthermore, are your data of sufficient quality and quantity?
Is your data biased?
It's not possible to eliminate bias in data, but you can be aware of it and work to mitigate it. You can reduce bias by incorporating multiple sources of data. And you can identify if you have any selection bias like survivorship bias that is easily overlooked.
Question?
Reach out to Nhan Ngo
nhan@seventyoneconsulting.se
