Provide frameworks that enable you to ask questions, because new risk emerges in a data-lead business
A safe assumption is that risk frameworks help identify and manage risk. However, if a risk framework is blind to new risk, asking questions does not unlock that the framework is blind.
If the quality of our data or the complexity of our model is far ahead of the other, we get poor quality decisions and worse outcomes. How do we use the risk framework to align the data quality, with the assessment of the quality of our models?
Our risk frameworks we built for a human in the loop, physical and process model. Complexity from data, outside dependencies within eco-systems, highly integrated data supply chains, delegated decision authority for mass automation, privacy, ethical processing, layered and tiered contractual relationship, fully automatous ML/AI are just an handful of risks that sit outside of traditional decision models and risk frameworks.
Humans move in and out of relationships and our mental and legal models worked within this conceptual framework. Digital contracts, tracking, consent notices, risks and use of data survives all users engagement. This is a change in the fabric.
How do we verify that the data being presented represents the problem and not a bias to a political gain or is insufficient to justify the outcome.
Human judgement, where a lack of data means that many outcomes are probable and possible, is the situation boards face. We build skilled teams with diversity of race, gender and thought to enable the best reasoned outcome within the risk framework we have agreed .
However the complexity of data and digital, coupled with speed means that our existing frameworks have bind spots, as this is all new. We have to develop the skills and new frameworks to cope with the new diversity of harm in a digital world.
The foundations of BI (business intelligence) and data governance are bed rocks of data for decisions. If we were still framed by the boundaries of our companies walls the tools we have created since 1990 to support decision making would be sufficient.