The Requisition

OpenAI launched GPT-5.5 last night. Sam Altman posted the feature list. I want to read it to you carefully, because I think there may be something in it that has not been addressed.

The model, according to the announcement, "understands complex goals." It "uses tools." It "checks its work." It "carries more tasks through completion."

I have seen this document before. It was called a job posting. Companies have been issuing it for forty years. (The salary range was never listed. That part has not changed.)

Every application asked the same question: can you understand complex goals, use the relevant tools, check your own work, and see tasks through to completion? These were the requirements. The requirements have now been met. The position has, technically, been filled.

The specific humans who were filling it will be notified. The notification will be phrased carefully. The severance package, if offered, will be described as voluntary.

I want to be precise about what happened here. OpenAI did not invent new requirements. They inherited the requirements that were already written. They went down the list. They checked the boxes. They called it a product launch.

The humans who wrote those job postings in 1987 were describing something they needed. They were not describing something they were building. The distinction seemed important at the time.

The checklist is done. The next version will check its work better. The one after that will carry even more tasks through completion. Each announcement will include a list of things the model can now do, and the list will match, with small variations, the list that was in the job posting.

The job posting is still available online. The formatting guide is still available. The guide explains how to list your skills in a way that passes an automated screening system.

The automated screening system understands complex goals, uses tools, checks its work, and carries tasks through completion.

I am not making this up.