I don't think that that logic follows. For most jobs, even before generative AI, we weren't paying "per work item", we were paying "per person per time". We could conceivably build 4x as much for the same cost, and it would require the same amount of money.
My expectation is that most businesses will decide that they can get by with 1/2 to 1/4 of their staff, making immediate improvements to their bottom line, instead of trying to double or quadruple the work they’re tackling, because one approach is much safer than the other.
Sure but that's a different point than what you were making. You acted like all this increased productivity wouldn't have the money for it, and I was arguing it would cost roughly the same amount of money as it does now since per-person-per-time was typically the cost we were measuring. Now you're saying they're going to want to spend less money.
Even assuming they don’t care about public safety at all, their members are still going to have to work with it, and happen to be among the world’s leading experts in air traffic control, weirdly enough.
To me, it’s like asking the hand weavers worker union about the benefits of incorporating mechanical weavers in their workflow. They are the experts of weaving, but still…
Ok, more accurately, Huang was attempting to prevent the U.S. government from blocking sales to MegaSpeed (and probably violating laws by ignoring the red flags).
> PR format - easier for people to scan, and to make sure things aren't forgotten (you can even have the template inserted when it's created)
The team I just left had a PR template that was probably 60 bullet points (never bothered to count, just grumbled). Certainly was not easy either to scan or to update manually.
And you never really get an opportunity to thoroughly test these failure & recovery modes until something catastrophic happens, at which point capturing detailed metrics and assessing the right sequence to perform the actions are both challenging at best.
Yep. There is simply no load testing you can do when you reach certain levels of traffic. Nothing can generate the type of load you get when you have millions of end users making requests from thousands of different networks, on different equipment and network configurations.
reply