You need to understand ROI to understand the value of AI

The people mocking Peter Steinberger's $1.3m AI bill keep comparing it to hiring six engineers for a year. He didn't have a year — he had 30 days, and that changes the maths completely.

It would appear that many of the people debating Peter Steinberger's use of AI, don't understand the concepts of false equivalence or human resource management.

We don't know exactly what Peter used these tokens for, so can't debate the work's value. To be nice (and make the maths easier), we will say that Peter used the tokens to successfully write 1,152,000 lines of code in 30 days. That's what he wanted, and that's what he got.

What's important, is how you compare costs for the AI and the human equivalent. Comments like the one from Linas Beliūnas that compare the $1.3m cost to hiring 6 US senior engineers for a year, don't make any sense.

We don't have a year here, mate. We need the code done in 30 days.

To get the same output in the same 30-day window, you’d need 72 senior engineers. Each writing 100 lines of code an hour, for 20 days (they get to have weekends off).

They all need to be hired, onboarded and brought up to speed with the project spec in 0 hours. Work flawlessly with 0 errors, no sick days, no management overhead, and then all fired on the last day without offboarding efforts (No goodbye cake either). We also need them to have their own laptops etc because there's no budget for that.

That's the real human equivalent of the AI effort.

And that's why AI is going to decimate employment in the tech industry.