Table 2.

Obstacles and challenges related to AI. Definition of explainability, uncertainty, and robustness.

	Explainability	Uncertainty	Robustness
Obstacle	The inability to monitor the mechanism of black boxes and correct the risk of unreasonable decisions leads to important ethical problems.	Uncertainty error is related to the use of raw data, which increases the amount of noise. Overfitting occurs when input data are not generalizable to the entire population and are more specific than a single location where they were collected.	Misinterpretation of misleading data that are misclassified.
Challenge	Explainable artificial intelligence would make the machine’s decision-making process known, allowing ethical problems to be overcome.	The quantification of uncertainty is crucial to increase confidence in the results obtained.	Robust model of correct recognition of contradictory input for accurate and correct classification.