|
|
|
|
|
|
by cracker_jacks
2243 days ago
|
|
|
The problem with these types of arguments with soft algorithms like machine learning is that the answer probably lies somewhere in between in the most literal way (it has non-zero influence from many input features). Accepting this ambiguity is where companies like Google/FB can get a free pass by blaming the automated algorithm. You could even create intentional situations where you know how the algorithm will behave because you know the data you are training on but still have the option to chalk up mistakes to 'accidental'. |
|
|