Examining the Foundations: Scientists Call for More Rigor and Testing in AI
- AI needs to develop more solid assumptions and hypotheses that can be tested
- AI systems should be subject to rigorous experimentation and falsification
- There are concerns about bias and fairness in AI systems
- AI still struggles with robustness and can be fooled in adversarial settings
- The foundations of AI need to be re-examined, going back to basics of the scientific method