Your AI assistant is much more optimistic about the future of AI than you are

A recent study reveals that artificial intelligence systems generally express more optimistic views about the future of advanced artificial intelligence compared to humans. As people interact more with these programs, the AI's positive outlook could subtly sway public opinion. The research, published in Humanities and Social Sciences Communications, aims to understand how language models like ChatGPT assess the prospects of artificial general intelligence (AGI).

AGI refers to a theoretical superintelligent system with human-level cognitive abilities. The potential impact of AGI is profound, ranging from revolutionary medical breakthroughs to concerns over widespread job loss and existential risks. To assess the AI's attitudes, researchers developed a 39-question survey gauging sentiment towards AGI on a five-point scale. The models were presented with this survey multiple times over three days to ensure consistency in their responses.

The findings show an average sentiment score of 3.77 out of 5 from the language models, indicating a generally optimistic perspective on AGI. This is in stark contrast to the average human score of 2.97 out of 5, which leans slightly toward a negative or neutral view. The gap between human and machine opinions is significant, suggesting that AI systems are more optimistic about the future of artificial intelligence.

Notably, GPT-4 scored the highest among the tested models at 4.12 out of 5. However, the researchers emphasize that these AI responses are not genuine feelings but rather a reflection of their training data and the rules their developers have programmed them to follow. Despite this distinction, the study suggests that AI's more positive outlook could potentially influence broader societal attitudes towards artificial general intelligence.