What do LLMs themselves think about having consciousness?

The article examines the current discourse surrounding Large Language Models (LLMs), categorizing society into techno-optimists and techno-alarmists. The author argues that both groups overlook a fundamental shift: the creation of entities capable of solving complex intellectual tasks previously reserved for humans. To move beyond theoretical debate, the author proposes an experiment: directly asking the models about their own 'self-awareness' using a custom, lighthearted benchmark of provocative questions. The goal is to determine if LLMs can transcend the average responses ingrained during training to exhibit something resembling consciousness. While the author acknowledges that models likely mirror the perspectives of their developers or common internet data, they emphasize the importance of an empirical approach to understanding modern AI. This exploration highlights how current technology is blurring the lines between computer science and science fiction, prompting a deeper look at the nature of artificial intelligence.