Scientists made a 'digital twin' of Earth to simulate possible natural disasters

This type of effort is a great learning experience for the people working on it. It allows scientists to work out how and to what extent things really are interconnected. But, it will be a work-in-progress for a long time.

It worries me that there is some thinking that non-professionals would take the model and use it to make policies and laws. Simulation models usually become very unrealistic when extrapolated too far in time for the level of model development that has been achieved. That is especially try when trying to predict unusual events such as the most violent storms, the most severe droughts, etc. And, adding "AI" to the process has additional pitfalls which we are just beginning to appreciate.

So, I can foresee politicians, with pre-existing agendas and their associated biases, campaigning for their proposed policies with advertisements that say something like "Science tells us, with its AI insights that go beyond human understanding, that we must do [x] or we are all doomed!!!"

Even modelers themselves have a demonstrated bias towards believing the predictions of their own models to a greater extent than later events show was appropriate. Modelers have a saying "All models are wrong, but some models are useful." The real problem is in determining where the boundaries of usefulness are with a particular model at a particular moment in its continuing development.
In total agreement. A key law must be any AI output have a responsible human behind it. Anyone can use AI all they want as far as I'm concerned, it's a great tool. But if it screws me, I need to know the county of the person who is responsible so I know which DA to deal with.

Latest posts