Load preset scenario
Run simulation
Presets
Import scenario
Export scenario
Main parameters
Additional parameters
Maximum fraction of automatable G&S tasks
AGI training requirements (FLOP with 2022 algorithms)
Effective FLOP gap (training)
Training goods vs R&D
AGI runtime requirements
Maximum trade-off
Labour substitution R&D
Returns to hardware
Returns to software
R&D parallelization penalty
Growth rate fraction GWP compute
Growth rate fraction compute training
Wake-up growth rate fraction of GWP buying compute
Wake-up growth rate fraction compute training AI models
Max fraction compute training
Initial biggest training run
Wakeup trigger
Effective FLOP gap (runtime)
Runtime goods vs R&D
Trade-off efficiency
Labour substitution goods
Substitution between capital and cognitive tasks for goods and services
Substitution between capital and cognitive tasks for hardware R&D
Research experiments substitution software
Efficiency of experiments for software R&D
Maximum hardware performance
Maximum software performance
Hardware adoption delay
Growth rate fraction capital hardware R&D
Growth rate fraction labour hardware R&D
Growth rate fraction compute hardware R&D
Growth rate fraction labour software R&D
Growth rate fraction compute software R&D
Wake-up growth rate fraction capital hardware R&D
Wake-up growth rate fraction labour hardware R&D
Wake-up growth rate fraction compute hardware R&D
Wake-up growth rate fraction labour software R&D
Wake-up growth rate fraction of compute software R&D
Max fraction capital hardware R&D
Max fraction of labour dedicated to hardware R&D
Max fraction of compute dedicated to hardware R&D
Max fraction labour software R&D
Max fraction compute software R&D
Max fraction GWP compute
Initial fraction capital hardware R&D
Initial fraction labour hardware R&D
Initial fraction compute hardware R&D
Initial fraction labour software R&D
Initial fraction compute software R&D
Initial vs cumulative input - hardware R&D
Initial vs cumulative input - software R&D
Initial hardware production
Accumulated hardware vs initial hardware production
Initial market hardware performance
Initial GWP
Initial world labour force.
Initial cognitive share goods
Initial cognitive share in hardware R&D
Initial compute share goods
Initial compute share R&D
Initial experiment share software R&D
Initial capital growth rate
Population growth rate
TFP growth rate
Compute depreciation rate
Money threshold training
Training requirements steepness (OOM)
Runtime requirements steepness (OOM)
Compute decomposition
Takeoff metrics
Show
the most important metrics
all metrics
Summary
Click
here
for the detailed year by year table.
Metrics