I bet you have seen a graph like this one before:
The most common implication from the posters and the most usual inference from the readers are that since the rollout of Large Language Models to the general public, information technology workers are hurt the most by it, as coding is one of the tasks that AI is best at. Coding jobs are easily replaceable by the LLMs and AI agents, the narrative goes.
I don’t have any comment on this narrative here except that I am sympathetic to it. It makes perfect sense to me also. The only problem is that it’s not the message the above graph says.
Movie Industry is the one hurt most
The sharp drop in headcount in the information sector is real, according to the US Bureau of Labor Statistics’ Establishment Survey.
As shown in the BLS webpage for the series (screenshot above), the payroll number peaked in November 2022, at a total of 3,115,000 people employed in the industry.
ChatGPT was released on Nov 30, 2022, so the two are “obviously” connected, right? Again, I am not saying the two things are not connected. The issue is we have to look at what the subcomponents of the information sector are.
BLS uses the North American Industry Classification System (NAICS) to arrange the industries in the payroll survey. Information is one of the so-called supersectors and these are the subsectors below them:
- Telecommunications
- Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services
- Web Search Portals, Libraries, Archives, and Other Information Services
- Publishing Industries
- Motion Picture and Sound Recording Industries
- Broadcasting and Content Providers
As you can see now, only the first three subsectors are what we would refer to as “information technology” industries. The rest are actually “media” industries.
And if we track the payroll decline in the information supersector with subsector decomposition, here is the picture:
Since Nov 2022, motion picture and sound recording, or what we commonly called “Hollywood,” suffered the largest job loss, with 126,300 jobs cut (as of Aug 2026). Telecommunications, part of the IT industry, followed right behind, shedding 96,500 jobs. Then it’s the publishing industry, which saw 82,700 jobs cut.
If we use the previous IT or media simple division to look at the situation, IT contributed 133,800 reduction in headcount while media industries cut 291,700 jobs.
Hence, the information sector job slump “since ChatGPT” was real and alive, but it was more a media industry problem rather than an IT industry problem.
If you want to find out more about BLS’s Establishment Survey industries and get the BLS series ID to build your own chart with Python and AI, check out our toolkit page, “BLS Payroll Survey Series ID Reference.”
