The research company focused on the long-term consequences of artificial intelligence (AI) usage, AI Impacts, has summarized the results of a large survey involving approximately 3,000 machine learning experts. They were asked to describe scenarios of how people's lives might change due to the implementation of neural networks. The study was published in the "Computers and Society" (CAS) section on the scientific publications portal arXive.
The majority of respondents expressed confidence that AI would bring about outstanding achievements in literature, mathematics, music, architecture, and various sciences.
However, between 38% and 51% of survey participants stated that, in their opinion, the likelihood of humanity's extinction due to neural networks is at least 10%. Approximately 60% of experts assessed such a possibility as at least 5%.
Addressing the threats posed by AI, specialists pointed out its ability to generate misinformation and influence public opinion, particularly concerning given the increasing authoritarian tendencies and decreasing freedoms worldwide.
Despite these concerns, 70% of research participants are optimistic about positive scenarios, basing their optimism on the growing capabilities of AI.
Some scientists believe that by 2047, with a 50% probability, machines will attain the ability to perform all possible tasks without human involvement and handle them better and with lower costs than humans. Two years ago, researchers thought such a development would not occur before 2060.