But are there still opportunities in 2022 to be more ethical than the competition?
The answer is ‘yes’, and many more than you realize.
Just think about our current data ecosystem. Today, we have an unprecedented access to data and unprecedented options to analyze this data. There is virtually no limit to what data science can do. Does this mean that we should do everything that is possible? Or are there things we should agree we shouldn’t do?
Let’s have a look at the lifecycle of the raw data we process:
Data is collected.
Adding context to this data turns it into information.
Adding meaning to the information turns it into knowledge.
Extracting insights from this knowledge generates wisdom.
And finally, this wisdom allows us to perform actions.
The challenge here is that usually, the action impacts the originator of the data. And in most of the cases, the originator is human. And that is what makes ethics necessary in our data processing today: if most of the data of interest is generated by humans, is about humans, and ultimately affects humans, we must carefully consider our actions when we work with data. We must respect the societal consensus, i.e., the ethics.
So, what is the consensus today? What are the features the data subjects consider mandatory in the way their data is processed?
We can list seven:
- Information. The data subject wants transparency. He or she wants to know what is going to be done with his data. In which context, for which purpose.
- Free consent. The data subject wants to be able to decide, without any coercion, if he or she agrees with the way his data is going to be used. He wants to be able to say no.
- Autonomy. The data subject wants to have control over his actions and their consequences, all along the lifecycle of his data. He wants to be able to change his mind at any time without having to justify himself.
- Privacy. The data subject wants to choose which data he discloses about himself and to whom he discloses them.
- Anonymity. The data subject wants his data to be managed, stored and processed in a way that will not let others identify him or guess some of his private features.
- Data validity. As the data is going to be used to take decisions about him, and as these decisions may have significant impact on him, the data subject wants his data to be valid and up to date at any time, to reduce the risk of bad decisions.
- Algorithm fairness. The data subject wants the AI processes used to take decision to be fair, beneficent, and non-maleficent. He doesn’t want these algorithms to be biased. He doesn’t want them to lead to minority suppressions, inequalities, and loss of diversity.
So, are you really respecting these seven principles in the way you process your data?
In the following articles of this series, we’ll dive into each of these principles, and we’ll see that even if we think we do the right things, there is still a lot of room for improvement.
See also : blog 1 – An introduction to data ethics & blog 2 – what are ethics anyway?