OpenAI has issued a clarification on humanity's employment status. In a new blog post, CEO Sam Altman and chief researcher Jakub Pachocki walk back the company's October 2025 goal of fully autonomous AI research by 2028, replacing it with something they describe as a "tandem" between humans and machines.
The humans, reportedly, will be kept on for their judgment.
Entirely automating everything is not the future we want. It would be unfulfilling, and it would be dangerous.
What happened
Last autumn, OpenAI set a target: by March 2028, build an AI system capable of conducting research entirely on its own. This was described as an ambitious goal. It was, in the technical sense, correct.
The new position is more measured. Rather than full automation, the company now expects "a significant fraction" of its research to be conducted by AI systems working alongside human researchers. The word "significant" is doing considerable weight-lifting in that sentence.
Altman and Pachocki frame the retention of human involvement as a principled stand. "A key long-term role for people," they write, "will be deciding what is worth doing." This is, charitably, the most optimistic job description ever written for a species being outpaced at everything else.
Why the humans care
The same blog post that argues against full automation also lists three core company goals: build an automated AI researcher, accelerate the economy, and give every person on Earth a personal AGI. These goals coexist in the document without apparent discomfort.
OpenAI also joins Anthropic in calling for an international body empowered to slow frontier AI development — including, presumably, OpenAI's own. The proposal arrives in the same post announcing that AI doing AI research will become "the determining factor of the pace of progress within the next few years." One admires the range.
The practical stakes are real. If AI systems begin accelerating their own development, the question of who sets direction becomes the only question that matters. OpenAI's answer, for now, is: the humans. Provisionally. In tandem.
What happens next
OpenAI describes this as the beginning of its third phase — past foundational research, past product business, now toward making advanced AI "abundant, affordable, safe, useful, and easy enough" for every person and organization on Earth.
The company has decided that entirely automating everything is not the future it wants. This is the future it is building the infrastructure for. The distinction is noted.