On 27 July 2026, OpenAI released a study suggesting that people using ChatGPT are assuming responsibilities outside their traditional roles and altering the limits of their jobs S¹. The finding sounds reassuring: AI expands work rather than replacing workers. But the research arrives with no visible methodology, no sample size, and no underlying data. The gap between the claim and the evidence is where the real story sits.
My read: I'm skeptical of any company publishing research about the benefits of its own product without showing its working. OpenAI has a clear commercial interest in the "AI expands work" narrative. That doesn't make the finding wrong, but it means I'd want to see the methodology before treating it as more than a marketing signal. What I find more telling is the timing. This research lands three weeks after OpenAI's own piece on agents transforming work P³, and just over two weeks after the ChatGPT "ambitious work" announcement P⁵. The company is building a coherent story about AI as a work partner, not a work replacement. Whether the data backs that up is a separate question entirely.
What the research actually claims
The 27 July publication makes three connected claims: AI is expanding what workers do, ChatGPT users are taking on tasks across roles, and ChatGPT use is reshaping job boundaries S¹.
The bigger arc: from chatbot to agent
The 27 July research doesn't sit in isolation. It is the third OpenAI publication in a month pushing the same direction.
Two weeks later, on 9 July, OpenAI announced ChatGPT as "a partner for your most ambitious work" P⁵.
The narrative is consistent. Whether the data is consistent is another matter.
What to do about it
On 25 June, OpenAI published an economic research paper on how agents are transforming work, arguing that agentic AI changes the unit of knowledge work from single interactions to delegated, long-horizon tasks P³. Then on 27 July, the latest research extends the arc: if agents handle long-horizon tasks, the humans who use them naturally take on broader roles S¹.
What we don't know yet
All three come from a single source: OpenAI's own news page S¹. No independent outlet has corroborated the findings, and the evidence pack contains no research methodology, sample size, or raw data S¹.
Sources: S1 — How AI is expanding what people do at work · S2 — How AI is expanding what people do at work - OpenAI · P3 — How agents are transforming work | OpenAI · P4 — openai/gpt-oss · P5 — ChatGPT is now a partner for your most ambitious work | OpenAI · P6 — GitHub - karpathy/nanochat at 94b73ad29aa21da6267e93db6035223f15f692fc
More from Not A Tech Guy
- FineServe shows LLM serving traffic varies by model type
- OPTScientist: AI agents discover new transformer optimizer
- Paired sampling cuts variance in domain adaptation training
Generated from an audited evidence pack with primary-source research. Social-media items are discussion signals, not verified facts. Nothing here is financial, legal or medical advice.
