Talks of AI token costs vs human labor is picking up steam. Brace yourselves.
3 replies (most recent on top)
@f7 the math is mathing here. Just about every AI (gpt / Gemini / Claude / Kimi) had very similar math - when prompting it to give conservative estimates on savings and productivity projections for implementation for a financial services company of our size.
Go look at your total compensation. Say your take home is $120k, $160k pre-tax. Chances are you’re total compensation (including insurance/ taxes / etc. that USAA has to cover) is more like $225k.
$225k x 1% of employees (340 employees cut)= $76 million / year.
$76M worth of tokens (for medium level tasks) is the equivalent of deploying out the productivity of 1,800 to 7,600 employees.
And that’s just with today’s models at our inadequate levels of data centers (why the token costs are high). It’s going to peak more then drop drastically in token costs. And already cutting 1% of employees replaces 1,800 to 7,600 worth of productivity.
Let that sink in.
What a d-mb post
Everyone is using AI to speed up their productivity right now
If anything the token cost is just going to slow us down