AI fundamentals · Question 3 of 60
Why are many deep learning models considered a "black box"?
1 answer is correct. Easy
Show answer
Answer
- C They can contain billions of parameters whose interplay is difficult for humans to interpret
Explanation
Modern deep learning models often have billions of parameters. A human can hardly read from them why they reach a particular result. In medicine, finance, defence or transport this is a problem, because decisions there must be justifiable. Legislators are therefore increasingly demanding transparency and explainability.