Policy level: influence on decisions requires an overview of impact and insight into consequences
Policymakers and other policy-level employees are increasingly expected to be able to 'get value from data', as it is called. There are various challenges to consider, because even at policy level people are used to thinking in abstractions. Moreover, policymakers are still quite far removed from the 'end user'.
And therein lurk dangers, such as oversimplification and (too) quick conclusions about large groups. Or just the tendency to create far too many, far too small boxes. Is that the basis for risk assessments, for example? Then that does not necessarily have desirable consequences.
Het Financieele Dagblad reported on the 'personalised' determination of insurance premiums and all the possible consequences that this has: exclusion of risk groups, higher or too low premiums for certain profiles, uninsurability and discrimination. Ultimately, it is of course the administrators who make the decisions about this, but as the most important 'influencers', policymakers have their own responsibility here. They should not only think in terms of possible benefits for south africa telegram data the system world. At this level, an overview of impact and insight into possible consequences for the living environment is necessary.
Risk assessment using algorithm (image: NOS.nl)
At the executive level, there is a desire to be pigeonholed and a fear of being pigeonholed
Finally, there is the implementation. I will zoom in on that, because here it is true: the closer to the customer, the greater the impact of decisions on individual people. Here too, there is a split issue, which essentially has two sides:
Pigeonholing
My colleagues and I see that account managers sometimes rely too quickly on the first image that emerges from the data. This can arise from the pressure of targets and administrative burden, which leads to the tendency to let time go before quality. The same behavior can result from the (unconscious!) desire to put people in pigeonholes, to be able to place them more easily. And then it can be difficult to deviate from the algorithm . In both cases, this is the big pitfall: a data profile does not provide a clear picture of someone.
Pigeonholing
We also see customer managers who are very suspicious of data (“my customers are so different, they can’t be pigeonholed”). This is often based on a worldview of ‘everyone.
