Source: statisticsbyjim.com
Certainly Yes!
There is a lot of talk about AI. Little is said about expert systems. The boundaries between one and the other are blurred and not easily discriminated.
An expert system arrives at the most probable solutions using algorithms that are validated in the “statistical” majority of cases.
AI learns from experience. In other words, he learns from statistics and from changing environments and from things that were not part of his experience before.
So statistics are crucial both in expert systems and in AI including what is called “generative” or that “seems” to generate/invent original things. In reality, it “only” exploited what it already knew before, enriching its knowledge and consequently modifying the algorithms that seemed valid until yesterday but which today need a new revision.
So without statistics none of this can happen.
The question then becomes: how deep does statistical knowledge need to be for the system to realize that the algorithms used are reliable to keep the risk at reasonable values?
Or why does the risk assessment calculated by humans differ from the risk calculated by AI predictive systems?
The solution lies in making all the decision-making powers aware of the understanding of the data: that is, of statistics applied to the goodness of the data.
More about this source textSource text required for additional translation information
Send feedback
Side panels
Last Updated on June 5, 2024
