Frontier AI IQ-equivalent
Benchmark-based IQ-equivalent trajectory for frontier machine intelligence.
Machine IQ-equivalent
Data: computer-iq-timeseries.csv computer-iq-timeseries.xlsx
Peak supercomputer performance
Rmax LINPACK performance in PFLOPS, shown on a log scale.
Rmax PFLOPS
Log scale.
Data: intelligence-rmax.csv intelligence-rmax.xlsx
AI autonomous task-duration milestones
Projected milestone dates by scenario, measured in equivalent human task hours.
HighMidLow
Log scale.
Data: ai-task-duration-projections.csv
Description
This module isolates the frontier capability evidence from the distribution comparison. It is about the leading edge.
Methodology
Charts combine benchmark-equivalent IQ trajectory, peak supercomputer Rmax, and autonomous task-duration milestone projections.
Data
All chart files for this module are local to this research folder.
Sources
- AI IQ Benchmark-mapped IQ-equivalent framing.
- TOP500 Supercomputer Rmax benchmark source.
- METR long-task horizon work Autonomous task-duration measurement context.
- OpenAI, Anthropic, and Google DeepMind model reports Frontier model benchmark anchors.
Limitations
Benchmarks do not measure wisdom, consciousness, virtue, or accountability. Task-duration projections are scenario models.
