Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
A recent study conducted by researchers at Stanford University, as reported by QuantaMagazine, has revealed that the perceived sudden advancements in the capabilities of Large Language Models (LLMs) are not as unexpected or unpredictable as previously thought. The study suggests that these perceived leaps in abilities are actually a result of the methodologies employed in measuring AI's performance. This insight challenges the common narrative surrounding AI development, emphasizing the importance of understanding the metrics used to assess AI capabilities.
"LLM capabilities are so surprising." You should be surprised by your own surprise at what statistical patterns can be drawn from huge datasets. As if such things could possibly be intuited in advance.
A new study suggests that sudden jumps in LLMs’ abilities are neither surprising nor unpredictable, but are actually the consequence of how we measure ability in AI. via @QuantaMagazine https://t.co/NWfSX6NTWF
LLM's are impressive and I was surprised by their abilities when I first saw them, but there seems to be a tendency to be credulous about AI and to attribute magical powers to these things, which I think fuels a lot of (utopian and dystopian) predictions about them https://t.co/K8nGcDJOzi
