Recent advances in AI have led to renewed optimism about computing's ability to help solve world problems, whether it's to cure disease, rein in climate change, improve education, or alleviate poverty. By now, however, we have witnessed many new digital technologies accompanied by promises to change the world for the better, only to fall short. Will AI be any different? In this talk, I suggest that technology's "Law of Amplification" provides an explanation for the mixed results of past technologies and a basis for predicting how AI might actually help -- and not help -- in the future. The law also leads to recommendations for anyone intent on ensuring that AI has positive impact.