Table 1.

Descriptive statistics of reported results over time for specific benchmarks and AI tasks

	NLP	Computer vision	Total
Benchmarks with ≥1 reported result	1318	2447	3765
Benchmarks with ≥3 results at different time points (% of above)	661 (50%)	1274 (52%)	1935 (51%)
AI tasks with ≥1 reported result	346	601	947
AI tasks with ≥3 results at different time points (% of above)	197 (57%)	386 (64%)	583 (62%)

A single task can be represented through several benchmarks.

NLP natural language processing.