From Geoff Rowe
Hal Hodson draws our attention to the increasing prevalence of automated systems that tend to be driven by predictive equations (7 February, p 30).
These equations often have one thing in common: they rely on incomplete and proxy information to make their predictions. It is sobering to consider how accurate similar predictive equations are in fields like medicine or epidemiology.
Although valuable insights are gained from epidemiological studies, say, the predictive equations derived from them may not be good enough to predict individual choice, behaviour or risk. This can be put down to the use of incomplete, inappropriate or inaccurate indicators.
The public should be given assurance that the predictive equations used in banking or national intelligence meet exceptionally high standards.
Ottawa, Canada
Advertisement
