"The blue 'theoretical capability' field represents outdated and heavily speculative educated guesses about where AI is likely to improve human productivity and not necessarily where it will take over for humans altogether."
"Anthropic's baseline for LLM 'theoretical capability' isn't based on empirical testing of its current models or quantifiable projections of performance increases over time."
A recent study suggests that LLMs could theoretically handle 80% of tasks in numerous job categories, including Arts, Media, and Management. However, this projection relies on outdated and speculative estimates rather than empirical evidence. The study's theoretical capability figures do not reflect current model performance or realistic productivity improvements. Instead, they stem from assumptions about AI's potential impact on human productivity, raising questions about the accuracy of these predictions regarding AI's future role in the labor market.
Read at Ars Technica
Unable to calculate read time
Collection
[
|
...
]