Data Science Wheel and Driving Ethical Behaviour
More organizations are pursuing a data-driven agenda, embracing AI and Data Science to achieve this ambition. But the ethical side of AI is complicated, and you may struggle to apply ethical considerations in your day-to-day work. This is where Intellerts’ Data Science Wheel can help guide organizations across the 5 different layers of ethical requirements of every AI project:
1. Data conditions
You may think good quality data is not important for good AI. You’d be wrong. Ask any experienced data scientist and they’ll tell you the same thing: to make correct (and therefore ethical) decisions based on your data, data quality is essential.
Another misconception is that data is objective. Bias within data, however, can lead to incorrect conclusions or reinforce existing prejudices within your data. As such, the state of your data and your data management efforts are vitally important. Data privacy and data security are, therefore, important boundary conditions for ethical data usage.
2. Model conditions
You must take data bias and quality also into account at the modeling stage. Bias can show up in the data and it can also be introduced when you select attributes for an AI model. The transparency of your model also matters and you must have justifiable reasons to opt for a more powerful but less transparent model. Transparency is not impossible to achieve. You can increase the transparency of, for example, a complicated neural network model by analyzing its operation or function, or by introducing human supervision. Either way, an AI model must be auditable to ensure the output of the model or the steps that led to the model can be replicated. An audit can be performed either internally or by an external company.
3. Data Scientist conditions
Whatever project you’re working on, it is unethical to act against your existing policies, rules or regulations. This tenet also applies to data scientists. But you must also have a clear agreement about accountability to ensure the consistency of your ethics across your team. Your data scientists must also adhere to proportionality and transparency, adopting the least intrusive data strategies and clearly documenting your policies, rules and regulations.
4. Impact on stakeholders
Employees or citizens may be impacted by the deployment of an AI and data science project. You should allow these employees to provide feedback across the project lifecycle, including after deployment. For citizens it should be possible to report (suspected) issues. Special considerations should also be made concerning the impact the data project on vulnerable groups, to prevent the project from having too great negative impact. Accessibility is another consideration and is all about allowing all people to have access to AI products and services. This will safeguard that certain groups within society will not be discriminated against when the application of AI based technologies advanced within our society.
5. Impact on community
From a social, environmental and democratic perspective, data projects must have a positive impact on our community. Cambridge Analytica’s use of Facebook data during the last US election is a clear example here of what not to do. You should also apply one final consideration: the headline check. If you cannot easily justify your data project in one simple sentence, you may want to leave it on the drawing board.
Integrated Approach of Data Science and Ethics
Now, enough about the theory. In practice, you may find it impossible to tick all of these ethical boxes. For example, the best performing AI model might be the least transparent one too. When this happens, you must carefully balance the pros and cons of your different project requirements against your ethical considerations, helping your data science teams achieve this delicate balance.
This is no easy task. That is why Intellerts has integrated these ethical requirements into its “The 8 steps of Data Science” data science methodology. This methodology outlines the necessary steps to successfully complete a data project which combines both the necessary methodological and ethical requirements. This way, ethics is applied in a practical way, to ensure the success of your data-driven agenda.
More about this: