"Explainable artificial intelligence" has the characteristics of open and transparent information. It will not only inform the people who interact with it what decisions it has made and what actions it is about to take, but belgium whatsapp number data 5 million also explain how it reaches the final calculation based on the existing data. result. "Explainable artificial intelligence" can perform the aforementioned actions while maintaining a high level of learning performance.
Current artificial intelligence imports data into machine learning programs to obtain the learned functions, but users are left with questions: Why does it do that? Why doesn't it decide otherwise? When will it succeed? When will it fail? Should I believe it? How do I fix the error?
In contrast, "explainable AI" uses new machine learning procedures with explainable interfaces to produce explainable models; such technology should be able to answer all of the above questions.
'At the same time, this is also accompanied by a certain degree of risk, because the decisions made by artificial intelligence are based on the data it imports. Although "explainable artificial intelligence" can enhance human trust in its decisions, if the data itself has If there are flaws, then such trust may also become some kind of "mistrust".
Another question is how well AI explains its decisions. An explanation is of no value if the user (who may be an average person without any technical background) cannot understand it. Scientists must work with user interface experts and incorporate complex mechanisms from explanatory psychology to try to think from the user's perspective to solve this problem.
