existup to dateIn the investigation,AI（AIExperts are alarmed by the breakneck pace of AI development, highlighting the divide in the tech world between accelerationists and doomsayers. The extensive survey of 2,778 AI researchers, conducted by San Francisco research firm AI Impacts, revealed a number of striking trends.
The survey results show that, overall,AI ExpertsThe trend of acceleration is being felt across the board. On average, the 50th percentile year for which AI-related tasks are expected to become feasible has moved forward 1.0 year, to between 2022 and 2023, across 32 different AI-related tasks.
However, one of the most striking findings is thatadvancedThe forecast time for machine intelligence (HLMI) and full automation of labor (FAOL) has shortened significantly between 2022 and 2023. Experts generally believe that the arrival time of HLMI has been brought forward by 13 years, while the forecast time for FAOL has been shortened by 48 years.
This remarkable shift in opinion shows that AI experts have come to believe that “for any occupation, machines can be built to perform the task better and cheaper than human workers.” This belief, which comes a century in advance, shows the optimism of experts about the pace of AI-driven automation in the world.
Although accelerationists hold their own, with 15.6% of experts believing AI will develop faster, completely eliminating the doomsayers (4.8%), the vast majority of experts remain on the sidelines. 29.9% of respondents believe AI will develop slightly slower, 26.9% believe it will be at current speed, and 22.8% believe it will be slightly faster. This mixed middle section accounts for 79.6% of all answers.
It is worth noting that the respondents in the survey only have expertise in artificial intelligence and may lack experience with non-technical factors that affect the trajectory of AI development. This requires us to be vigilant when reading experts' views on the future.
Experts are full of expectations for the future of artificial intelligence, but some are also concerned about this accelerating technology train. With the rapid development of AI, the scientific and technological community is divided, and experts seem to be more inclined to the acceleration side. This survey result not only reflects the views of artificial intelligence experts, but also is an important guide for the development of this dynamic field.