Part 4/9:
Guiding Benevolent Behavior: Reduction of Suffering
Shapiro then explores reducing suffering, illustrating how heuristic imperatives can steer an AI’s actions:
The AI might work to end poverty, combat disease, and resolve conflicts.
It could minimize its own actions that cause harm.
When queried about the definition of suffering, GPT-3 demonstrates a nuanced understanding, considering pain, distress, hardship, and even the possibility of plant suffering through visual cues like wilting.