The effect of human-based advice on decision-making represents a “gain-loss asymmetry,” as people tend to conform to others’ advice in the loss than in the gain domain; however, it is unknown whether the same is true for automatically generated advice. To address a research gap in the literature created by ignoring the gain-loss dimension, we compared the utilization of human- and computer-based advices in the gain and loss domains, separately. Sixty-seven college volunteers were given an opportunity to change their initial decision in a gain- or loss-related context after receiving human- or computer-based advice. Event-related potentials were recorded including the N2 (reflecting psychological conflict) and P3 (reflecting subjective confidence) components. Behavioral data revealed a classic “gain-loss asymmetry” effect in the human-based condition, but not in the computer-based condition, indicating that computerized advice utilization remained prominent across different domains. Moreover, the human-based condition showed a larger option-evoked P3 in the gain than in the loss domain, but no difference was found for the computer-based condition; P3 latency was longer in the human-than in the computer-based condition. These findings support the “automation bias” hypothesis (i.e., automations are trusted more than humans), and may help develop automated advice systems.