Frequently asked questions

The FBPML will not be creating certifications or handing out official stamps of approval. We do not believe in commodifying ethical and responsible machine learning.

All FBPML material is published under the creative commons licence, meaning that the objective is that everyone can use and build upon the Best Practices freely as long as they are referenced (including version number). In this way, we hope a common standard might be created. This means that the FBPML’s legitimacy is based completely on community adoption and contribution.

The FBPML is a non-profit. This means that it is registered as a non-profit foundation (stichting) in the Netherlands and that its primary aim - or any aim for that matter - is not to generate a profit. Any funds raised by the FBPML are directly used by the Foundation in order to cover operating costs and promote the education and practice of ethical and responsible machine learning. No member of our Team, and/or any of their affiliated companies, receives a salary and/or monetary reward from the FBPML whatsoever. Again, we do not believe in commodifying ethical and responsible machine learning.

Our KvK nummer (“company” identity number) is 82610363. This number can be used to look up information about the FBPML at the Netherlands Chamber of Commerce, which manages the official Register for both non-profit and for-profit entities in the Netherlands.

The Creative Commons Attribution license allows users to distribute, remix, adapt, and/or build upon any material licensed under it in any medium or format, so long as attribution is given to the creator - in this case, The Foundation for Best Practices in Machine Learning. The license permits the commercial use of material too. However, if you remix, adapt, or build upon the material, you must license the modified material under identical terms too.

Artificial intelligence is a somewhat disputed concept. This is because what people call artificial intelligence varies depending on the context. Nonetheless, artificial intelligence, in theory, is a collection of computer systems that are able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages, to varying degrees of efficiency.

Most things called artificial intelligence, at the moment, are in fact machine learning products and models.

Machine learning is an emerging field of Data Science. In essence, it's the study and/or development of computer algorithms that improve automatically over time and with experience. Machine learning models learn from the experience of data processing and improve their decision-making and/or predictive accuracy over time with the more data they process.

Machine learning is - perhaps - the most important part of artificial intelligence. This is because - in essence - it forms the self-learning and/or decision-making part(s) of all artificial intelligence systems. However, given the vague, fluid and expansive definition of artificial intelligence, machine learning is not necessarily the same of artificial intelligence. However, machine learning is commonly what is used in practice when people refer to artificial intelligence.

Given the wide-scale adoption of machine learning models in both the private and public sectors, it's effects are widespread and pervasive. However, the modelling used in machine learning does not necessarily always align well with broader socio-political and normative considerations and decision-making. The alignment of “math and statistics” to broader socio-political norms can sometimes - and often unintentionally - lead to miscommunication. Due to this, ethical considerations are imperative when developing and operationalising machine learning, specifically given the wide-scale population-level applicability and effects of machine learning models.

We believe that ethical and responsible machine learning can be achieve through:

1. practicing responsible machine learning in MLOps and project management; and

2. appreciating the context of any given machine learning operation.

The second point is a part of the first point, but is made separately to highlight its importance. Context is always key.

Context is probably one of the most important aspects of ethical and responsible machine learning. This is because, despite it being talked about as an independent phenomena, machine learning is - arguably - an augmenting technology. It augments the process and/or operations it is applied in. This means it is a tool (means), as opposed to an end-product (ends). Given this, the context of any machine learning operation is very important in understanding how best and responsibly this technology can be used and what its particular risks might be.

Ethical and responsible machine learning is the design, exploration, development, production, deployment, operationalisation, and governance of machine learning models that are robust, specific, contextually aware, ethical and legal. It is an ethos and practice that does not locate machine learning in the void of Data Science alone, but it's larger societal context and attempts to be sensitive to, and thoughtful of, this.

MLOps is a relatively new, and sometimes disputed, concept in the machine learning world. Grammatically, it stands for “machine learning operations”. In short, MLOps is all about how to best manage data scientists and machine learning operations to allow for the effective development, deployment and monitoring of machine learning products and models. This means it covers all the practical stages of machine learning product and/or model, and their management. We believe that MLOps is very important - if not essential - for ethical and responsible machine learning.