2024 ThousandsofAIAuthorsontheFuture

From GM-RKB
(Redirected from Grace et al., 2024)
Jump to navigation Jump to search

Subject Headings: AGI Prediction, Automation of Human Occupations, AI Risk, Scientific Progress.

Notes

  • A 50% probability by 2028 for AI systems to achieve several milestones, such as:
    1. Autonomously constructing a payment processing site from scratch.
    2. Creating a song indistinguishable from a new song by a popular musician.
    3. Autonomously downloading and fine-tuning a large language model.
  • If scientific progress continues without disruption, there's a:
    • 10% chance by 2027 and a 50% chance by 2047 that autonomous intelligent machines will outperform humans in every possible task.
    • This 2047 estimate is 13 years earlier than the one reached in a similar survey conducted in 2022.
  • The probability of all human occupations becoming fully automatable is forecasted to:
    • Reach 10% by 2037 and 50% by 2116.
    • This is earlier than the 2164 forecast made in the 2022 survey.
  • Regarding the long-term value of AI progress:
    • 68.3% of respondents believed that positive outcomes from superhuman AI are more likely than negative ones.
    • 48% of these optimists assigned at least a 5% probability to extremely negative outcomes, including human extinction.
    • 59% of pessimists gave a 5% or higher chance to extremely positive outcomes.
    • 38% to 51% of respondents assigned at least a 10% chance to advanced AI leading to outcomes as severe as human extinction.
  • Over half of the survey participants indicated that "substantial" or "extreme" concern is warranted regarding six different AI-related scenarios, including:
  • Lack of consensus on whether faster or slower AI progress would be more beneficial for humanity, but:
    • Broad agreement on the need for more research aimed at minimizing potential risks from AI systems.

Cited By

Quotes

Abstract

In the largest survey of its kind, 2,778 researchers who had published in top-tier artificial intelligence (AI) venues gave predictions on the pace of AI progress and the nature and impacts of advanced AI systems The aggregate forecasts give at least a 50% chance of AI systems achieving several milestones by 2028, including autonomously constructing a payment processing site from scratch, creating a song indistinguishable from a new song by a popular musician, and autonomously downloading and fine-tuning a large language model. If science continues undisrupted, the chance of unaided machines outperforming humans in every possible task was estimated at 10% by 2027, and 50% by 2047. The latter estimate is 13 years earlier than that reached in a similar survey we conducted only one year earlier [Grace et al., 2022]. However, the chance of all human occupations becoming fully automatable was forecast to reach 10% by 2037, and 50% as late as 2116 (compared to 2164 in the 2022 survey). Most respondents expressed substantial uncertainty about the long-term value of AI progress: While 68.3% thought good outcomes from superhuman AI are more likely than bad, of these net optimists 48% gave at least a 5% chance of extremely bad outcomes such as human extinction, and 59% of net pessimists gave 5% or more to extremely good outcomes. Between 38% and 51% of respondents gave at least a 10% chance to advanced AI leading to outcomes as bad as human extinction. More than half suggested that "substantial" or "extreme" concern is warranted about six different AI-related scenarios, including misinformation, authoritarian control, and inequality. There was disagreement about whether faster or slower AI progress would be better for the future of humanity. However, there was broad agreement that research aimed at minimizing potential risks from AI systems ought to be prioritized more.

References

;

 AuthorvolumeDate ValuetitletypejournaltitleUrldoinoteyear
2024 ThousandsofAIAuthorsontheFutureKatja Grace
Harlan Stewart
Stephen Thomas
Ben Weinstein-Raun
Jan Brauner
Julia Fabienne Sandkühler
Thousands of AI Authors on the Future of AI2024