MemCast
MemCast / episode / insight
#ai-adoption21#economic-impact5#employee-productivity4
Current AI provides ~15-20% productivity gains in coding
  • Estimates coding productivity improvements have grown from ~5% to 15-20% in six months
  • Predicts this will accelerate to 40%+ soon
  • Even small % gains matter at scale given software's economic importance
  • Full automation of SWE tasks would require solving 'closing the loop' problems
Dario AmodeiDwarkesh Patel00:37:51

Supporting quotes

Quote
00:37:51
“I would say right now the coding models give maybe, I don't know, a 15-20% total factor speed up. That's my view. Six months ago, it was maybe 5%.” — Dario Amodei
Quote
00:39:13
“As you go, Amdahl's law, you have to get all the things that are preventing you from closing the loop out of the way.” — Dario Amodei

From this concept

Economic Diffusion vs Technical Progress

Amodei distinguishes between the exponential of technical capability and the (still fast but slower) diffusion into the economy. He discusses why adoption isn't instantaneous even with transformative AI.

View full episode →

Similar insights

“AI will control roughly 70 % of the global economy by 2040”

AbhigyaDhruv Gupta

“AI will reshape food security, potentially causing famine around 2040”

AbhigyaDhruv Gupta

“Countries that fail to integrate AI will fall behind in the upcoming geopolitical hierarchy”

AbhigyaDhruv Gupta