Hi ,
One of the first lessons I learned while working as a university lecturer is that it's OK to say "I don't know."
Nobody expects anyone to know everything - even an academic who
presents themself as an expert in their domain.
In fact, pretending to know the answer when you don't can often do more harm than good. It's better to admit you don't know, and then go off and find the right answer, than to say the wrong thing and risk leading people astray.
This approach also serves to boost your credibility. Because if people learn you will only answer if you have a high level of confidence in your response, they will trust your answers even more.
Here's the Thing...
While humans have the option of admitting they don't know something, the models we build as data scientists rarely do.
If you were just 50.1% certain that option B was better than option A, there's a good chance you would hold off on deciding until you had more information at hand. Yet we design our models such that, in the same situation, they are forced to
choose option B.
AI aims to replicate human intelligence, but one of the most intelligent things a human can do is to know when to admit they don't know.
If you want to build a smarter model, give it "permission" to not know.
Talk again soon,
Dr Genevieve Hayes.
