A recent survey of 475 AI researchers indicates that 76% believe scaling up existing AI methods won’t lead to artificial general intelligence (AGI). This finding challenges the tech industry’s prevalent strategy of enhancing generative models by increasing hardware. Many researchers suggest that the focus on scaling has plateaued, stressing the importance of understanding AI advancements. Additionally, the heavy investments in AI infrastructure, particularly in generative AI, raise concerns about energy demands, as illustrated by Microsoft's plans to power its data centers with nuclear energy.
An overwhelming 76 percent of AI researchers believe that simply scaling up current AI approaches is unlikely to lead to achieving artificial general intelligence.
Stuart Russel asserts that the previous focus on scaling without understanding has plateaued, indicating a shift in perceptions among AI experts.
The substantial investments in AI infrastructure, with venture capital in generative AI surpassing $56 billion, point to a widespread concern over sustainability.
Microsoft's significant commitment to AI infrastructure financing illustrates the ongoing arms race, coupled with unprecedented energy demands due to AI hardware.
Collection
[
|
...
]