LifeVal is a new benchmark created from feedback from thousands of ChatGPT users to target everyday assistant tasks. It is being used to measure and improve model performance on tasks such as childcare, government services, and personal finances.
Key points
LifeVal benchmark shaped by feedback from thousands of ChatGPT users.
Used over the past year to measure and improve model handling of everyday tasks.
Sources
- Original post
- openai.com - Linked in the post
Anthropic S-1 adds claim imminent RSI danger threshold
Anthropic’s S-1 filing adds new language on page 29 saying internal estimates indicate they may imminently meet the RSI danger threshold of dramatically accelerated progress. The claim describes progress in one year as equal to two years of the 2018–2024 baseline.
Key points
Page 29 of Anthropic’s S-1/A adds new language about internal estimates.
Internal estimates indicate they may imminently meet the RSI danger threshold, defined as one year of progress equal to two years of the 2018–2024 baseline.
