News: 1769449020

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

AI adoption at work flatlined in Q4, says Gallup

(2026/01/26)


AI adoption in the workplace stalled in the fourth quarter of 2025, but those who have already started using it are making increased use of it, according to a survey by pollster Gallup. Don't let that fool you into thinking AI is taking over work, though: frequent AI users are still a tiny minority of overall workers.

Sharp increases in the workplace usage of AI throughout the year almost entirely leveled off between Q3 and Q4 2025, Gallup [1]reported over the weekend. The total percentage of workers using AI climbed just a single percentage point in the final three months of the year, from 45 percent in Q3 to 46 percent. Similarly, respondents noting that their company had adopted AI hovered at 38 percent in Q4 – just a single percentage point higher than Q3 adoption [2]numbers .

That said, while AI adoption and usage were flat over the final quarter of 2025, the number of workers reporting frequent and daily usage of AI did increase.

[3]

There was a three percent increase in frequent AI usage, defined as being a few times a week, and a two percent increase in daily AI usage. While greater than the overall one percent increase, those changes are still small, and it's also worth noting they comprise a vast minority of AI users overall. Per the responses, only 26 percent of workers are using AI frequently, and just 12 percent are using it daily.

[4]

[5]

So who's jumping aboard? The results there aren't surprising either.

Per Gallup, knowledge-based industries are dominating in AI usage, with the tech sector particularly fond of it – 77 percent of tech professionals report using AI in the workplace, with 57 percent doing so frequently and 31 percent using AI daily. University workers and finance professionals are also high adopters.

[6]

Interestingly enough, workers in "remote-capable" roles, which Gallup defines as jobs that "could reasonably be completed remotely regardless of where the employee actually works," are big fans of AI, with 66 percent of them reporting using it in Q4. Far fewer employees in non-remote capable positions report using AI – just 32 percent – though that could be the nature of their jobs, as Gallup notes remote positions tend to be desk and office-based, while on-site jobs often involve more physical work.

Leadership positions, which Gallup defines separately from managers and individual contributors and which tend to be desk- and office-based roles, are also more likely to use AI than other workers, the pollster said, while noting the gap between leadership and rest-of-the-workplace AI usage has widened in recent years.

[7]One in six US workers pretends to use AI to please the bosses

[8]One long sentence is all it takes to make LLMs misbehave

[9]McKinsey wonders how to sell AI apps with no measurable benefits

[10]AI pilots keep crashing, mostly because firms skip the prep, survey finds

Taken as a whole, Gallup argues that the growing divide between users and non-users, and stagnating expansion, means that AI has a use-case problem.

"Gallup research shows that lack of utility is the most common barrier to individual AI use, suggesting that clear AI use cases may be more apparent for leaders than employees in other roles," the polling agency concluded of its Q4 numbers.

"This underscores the importance of grounding decisions about AI adoption in a clear understanding of how AI may be applied to different roles and functions, not just among those closest to decision-making," Gallup added.

[11]

In other words, the C-suite ought to start thinking outside its own experience on AI adoption, because the average employee [12]doesn't seem to find it particularly useful . That, or they're realizing AI [13]isn't saving them any time , and may even be [14]eating up more of it . ®

Get our [15]Tech Resources



[1] https://www.gallup.com/workplace/701195/frequent-workplace-continued-rise.aspx

[2] https://www.gallup.com/workplace/699689/ai-use-at-work-rises.aspx

[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aXfyIQAQanmuuJtwtrKQLwAAAZc&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXfyIQAQanmuuJtwtrKQLwAAAZc&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXfyIQAQanmuuJtwtrKQLwAAAZc&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXfyIQAQanmuuJtwtrKQLwAAAZc&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[7] https://www.theregister.com/2025/07/22/ai_anxiety_us_workers/

[8] https://www.theregister.com/2025/08/26/breaking_llms_for_fun/

[9] https://www.theregister.com/2025/10/09/mckinsey_ai_monetization/

[10] https://www.theregister.com/2025/11/13/survey_ai_projects_failure_no_prepare/

[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXfyIQAQanmuuJtwtrKQLwAAAZc&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[12] https://www.theregister.com/2025/11/19/ai_force_feeding/

[13] https://www.theregister.com/2025/12/24/reason_ai_isnt_delivering/

[14] https://www.theregister.com/2025/10/21/ai_eats_leisure_time/

[15] https://whitepapers.theregister.com/



3rd story today which says LLM code assistants aren't all that

Dan 55

Did anyone think to stop and ask the developers? We could have told you that for free.

There'll be plenty of work soon for programmers who know how to program instead of prompt, they'll have to roll back LLM-generated code blindly pushed to main and do the job properly.

Re: 3rd story today which says LLM code assistants aren't all that

Pickle Rick

I wonder how much of the (global) COBOL code base is now FUBAR thanks to "pros" using LLMs. Dig out those code sheets girls and boys! >cha-ching<

Joking(?) aside, rolling back code should be the least of their problems[1], unborking the data might be the PITA. >cha-ching cha-ching<

[1] You did back up your code and keep CVS/git repos up to date, right? Riiiight?!

Re: 3rd story today which says LLM code assistants aren't all that

elsergiovolador

LLM can produce robust, performant, production quality code, but it needs a lot of nudging and back and forth. It also can save a lot of time, but if someone cannot recognise good from the bad, god help them.

Re: 3rd story today which says LLM code assistants aren't all that

Dan 55

It can produce robust, performant, production quality code if it's a small program which fits in its context window. Feeding it hundreds or thousands of files is doomed to failure.

So if you spoonfeed an LLM chatbot little bits of code it can give an answer which solves the problem you have, if you don't want to think about programming.

If you use a coding agent which pulls in all the code it runs the risk of going off the rails. If not, it will probably just jam more code in there instead of refactoring. And you will probably not see how bad all the code is just from the pull requests you get from the agent before approving them.

"frequent AI users are still a tiny minority of overall workers"

zimzam

And how many of those are mandated users?

"...means that AI has a use-case problem."

Pickle Rick

Translation: all AI LLM/GAN use cases are identified and saturated.

NB: most trials to confirm potential use cases yielded poor results and are advisedly to be discontinued.

YMMV

Anonymous Coward

Define "adopted". Does that include Micro$oft pinning you down and forcefully shovelling CoPillock down your throat?

"the C-suite ought to start thinking outside its own experience"

Michael Hoffmann

Could just about start and end anything with that statement!

When somebody tells me...

JacobZ

...that AI helped them with their work, it doesn't raise my opinion of AI. It lowers my opinion of their work.

"Whoever undertakes to set himself up as a judge of Truth and Knowledge is
shipwrecked by the laughter of the gods."
-- Albert Einstein