Right steps should be clearly specified. Then only the problem of Facebook algorithm and other algorithm will be solved. Let us make AI white box AI no more black box AI. These are the following points from me about the white box AI.
- White box means transparency- White box AI refers to a phrase which means that there should be transparency in the input and output the code produces.
- Just like white box testing, one should be able to fill in different hardcoded inputs to see the results that will be obtained.
- The current AI is black box AI no one knows what the real output will be from a given input.
- Further the code in-depth analysis is not made.
- Even software’s are tested for Black Box testing- which means the testes who don’t know coding can still check the input with the expected outputs
- Same for Black Box based AI — the people who don’t know data science can at least do Black Box AI testing — to see what inputs are given to the AI based software and whether output generated is appropriate, if not same all time.
- And those who know data science can do the Red-Box testing- meaning , varying the parameters to do the testing and for efficient parameter learning.
- Further, those who know how models’ were implemented need to do white box testing.
- The testing should be such that right results are obtained
- These are basic steps any graduate learns in computer science — How to do testing, white box testing and black box testing, why AI based software’s not doing the efforts ?
- Are they just making an algorithm and putting it to use live?
- We need transparency in input output –even if it is NLP.