,
3 tweets,
1 min read
Read on Twitter
AI has managed to follow roughly linear progress over the past 70 years thanks to an exponential increase in available resources (computing power, researcher headcount, tooling/networking improvements).
Like in other successful scientific fields, the exact rate of exponential resource consumption seems to adjust itself so as to sustain this roughly linear progress. An interesting game of incentives. It's not unlike the Bitcoin mining market.
I think this exponential increase in resources for AI research can be sustained for a few more decades. Meaning that we'll continue to see linear progress at roughly the current pace for a while.
