Boyd's OODA loops and Kahneman's Fast System 1 thinking resemble how a narrow AI conducts inference. One pixel attacks fool AI's. Implications for healthcare discussed.