Amodei named why the pace changed

Something you bought this year has already changed under you. The tool works differently, or it costs differently, or what your team learned to do with it is now only half right.

You have been reading that as bad luck, or as a vendor being sloppy. It is neither.

The reason is in Dario Amodei's post and almost nobody quoted it. He does not only say the pace is fast. He says what changed and when: "since roughly this summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI."

That is compounding, which is a different problem from fast.

What got covered was the row around it. Per Axios, on 14 September Trump called the fear a hoax and said a strong, smart president is the only guardrail AI needs. China's foreign ministry called the calls to slow down fear-mongering. Both inside a week.

The same Axios piece records that Amodei, OpenAI's Sam Altman and Elon Musk all agreed AI is becoming more powerful than humans can understand and monitor. Three of the people building it. Not one of them slowed.

Nobody is slowing down, so nothing you buy settles, and every hour the automation saves you turns up as work somewhere you were not counting.

I can only tell you what that has done to us, and we are the company we try everything on first.

At one point our API bill got really high, and we were producing more content than the pipeline could handle, so we changed the way we worked.

We built an automation for thumbnails that cost more time than it saved. We use something that produces B-roll automatically, and you still sit and edit it by hand.

And we stopped reaching for the biggest model unless the job needed it, because it would spin up far more than the work required when a simpler setup did it better.

None of that is a vendor being sloppy. That is what buying into something unsettled feels like from the inside.

The saving is real, and it moves, and nobody goes back and updates the business case when it does.

Staying ahead of that is a full-time practice. You have to work in it, build with it, follow it daily. That is not your job, it is not going to become your job, and the pace is not waiting for you to find the time.

So, the pace is not yours to set. What is yours is where the skill sits. If it sits in your people, it survives the next update. If it sits in a setup someone built for you, it might not.