Employers that use AI and algorithmic decision-making tools must be careful that the technology does not systematically disadvantage people based on their race, color, religion, sex or national origin, according to a new guidance document from the Equal Employment Opportunity Commission (EEOC).