Researchers Record Their Work So AI Can Do Their Jobs

Before we get to today’s main item, please check out my colleague Sri’s exclusive piece from Friday night revealing that OpenAI expects to burn $115 billion in cash through 2029, plus tons of other financial data the company has disclosed to investors. 

We are still having a hard time digesting the capital that cash burn will require the company to raise in the next five years! We will bring you more information and analysis on it later in the week.

Now onto today’s column…

AI researchers are training their own replacements. Yes, they’re taking seriously the possibility that AI models could one day perform nearly any digital task better than a human. With that in mind, some researchers are recording their work in the hopes that AI models will someday use the data to learn to do their jobs, which could help speed up their research. 

If future AI can act like a coworker, these researchers say, what better way to onboard an AI employee than to have them read transcripts or hear recordings of every meeting an organization has had, and see every computer-based decision that humans in the organization have made?