Company-reported retrospective task shares

Two zero-baseline bars show OpenAI's rounded retrospective percentages. Previously unsolved tasks were 22 percent of 898 tasks, while 93 percent of board-discussed tasks came from that set. The measures have different denominators.

OpenAI reports that a previously unsolved task set represented 22 percent of all tasks and 93 percent of board-discussed tasks.

Semantic data for Company-reported retrospective task shares
RecordValueUnitsState
Board-discussed tasks from that set | separate denominator93company-reported percentCaution
Previously unsolved tasks | 198 of 89822company-reported percentCaution
Company-reported retrospective task shares Two zero-baseline bars show OpenAI's rounded retrospective percentages. Previously unsolved tasks were 22 percent of 898 tasks, while 93 percent of board-discussed tasks came from that set. The measures have different denominators. OpenAI reports that a previously unsolved task set represented 22 percent of all tasks and 93 percent of board-discussed tasks. VALUE / company-reported percent Board-discussed tasks from that set | separate denominator 93 company-reported percent Previously unsolved tasks | 198 of 898 22 company-reported percent
Sources
Retrieved
Units
company-reported percent
Transformations
  • Do not divide the values, calculate uplift, or imply causation.
  • Keep the 198 of 898 count and the separate board-discussion denominator in the labels.
  • Render both rounded percentages on a zero-to-100 scale.
Uncertainty
OpenAI reports rounded percentages. Public counts behind the 93 percent value, uncertainty intervals, and independent replication are unavailable in the reviewed record.
What this figure does not prove
There is no causal implication, common denominator, uplift measure, settled intent, production-wide result, or independent replication.