From the archive
Friday, 6 November
Folded from 14 sources that day.
Sorted by quality and by how many sources covered it, the same for everyone.
-
Measuring how AI agents handle everyday wet-lab science
OpenAI's BioBasicsBench is a 100-task benchmark testing how well AI agents interpret raw experimental data and handle routine wet-lab work.
-
What Huawei's Horinger data center reveals about China's compute gap
Huawei's Horinger AI campus in Inner Mongolia has under a tenth of Colossus 2's capacity. Its Ascend 910C delivers about 0.76 H100-equivalents, and its efficiency gap limits what China's power can support.
-
Egyptian and Nigerian English join the Open ASR Leaderboard
The Open ASR Leaderboard adds private Egyptian and Nigerian English splits from Liva AI. Its accent breakdown now groups results by Kachru's Three Circles.