Models in EA-Funds

/*
extremely quick model to get a sense of number of possible future digital minds without explosive growth relative to current levels
*/

flops_required_to_sim_brain_per_sec = 1e15 to 1e21
dollar_per_current_flop_per_sec = 1/exp(10)
current_cost = flops_required_to_sim_brain_per_sec * dollar_per_current_flop_per_sec

cost_of_brain_sim(t) = current_cost* (0.5^(t/2.46))

/*
Describe your code here
*/

a = normal(2, 5)

/*
Estimating population of Japan in 1600
*/

global_population_today = normal(8.1, 0.2)

japan_pop_today = normal(0.125, 0.005)

ratio = japan_pop_today/global_population_today

/*
Describe your code here
*/

/*

I mean, at a high level how am I thinking about this stuff? Seems like the best case here is something like "well, this does seem like a really fun way to learn about AI Safety, and could become one of the default intros". 

However, I do think this is mostly capability-progress bottlenecked. But I also feel like I would want someone to work on this, since it does seem like it is the kind of thing that could be really great as soon as it becomes feasible. 

/*
Describe your code here
*/

a = normal(2, 5)

/*
Describe your code here
*/

a = normal(2, 5)

/*
Describe your code here
*/

a = normal(2, 5)