AI Use at Work Nearly Doubles in Two Years (gallup.com) 31
AI use among U.S. workers has nearly doubled over two years, with 40% of employees now using artificial intelligence tools at least a few times annually, up from 21% in 2023, according to new Gallup research.
Daily AI usage has doubled in the past year alone, jumping from 4% to 8% of workers. The growth concentrates heavily among white-collar employees, where 27% report frequent AI use compared to just 9% of production and front-line workers.
Daily AI usage has doubled in the past year alone, jumping from 4% to 8% of workers. The growth concentrates heavily among white-collar employees, where 27% report frequent AI use compared to just 9% of production and front-line workers.
google (Score:4, Informative)
Re:google (Score:4, Interesting)
generating exec summaries, seems to be the best use case as no one actually reads them.
Bullshit task for the bullshit generator (Score:2)
That fits.
Re: (Score:1)
Mod grandparent funny, though the parent Subject seems better for the joke? Generally misfiring of the Funny on Slashdot these years?
On the story, I think the "use" should be in the more active sense, but 'I don't have to work no more' [sic], so that part of the story flies over my head, but I sometimes find AI a useful tool and use it accordingly. About two days ago an ancient website crashed and I couldn't get some information I wanted. So I used an AI to quickly create a little tool to generate the infor
Re: (Score:2)
generating exec summaries, seems to be the best use case as no one actually reads them.
Pretty much 50% of my AI use is generating useless guff that no-one reads. I know they don't read it because they call/teams me 10 minutes after getting the ticket to have me explain it to them in simple words with fewer than 3 syllables. If the lazy fuckers aren't going to read it, I'm not going to bother writing it.
Re: (Score:1)
No, "included in your Google Search" is not what they mean.
Gemini and ChatGPT's MAU numbers are all you need to stop saying stupid shit going forward.
Re: (Score:2)
There is a difference between triggering it and using it. If you actually use the summary in Google search it counts, if you just keep scrolling it doesn't.
Power consumption? (Score:2)
We know how expensive AI s, resource-wise. I wonder if the power consumption is on the same track.
Re: (Score:1)
The argument becomes more and more irrelevant. Ressource usage is going down by a factor of 10 per year. That means factor 1000 in the last 3 years.
Here is a logarithmic plot, which by now even has a few newer models missing: https://d1lamhf6l6yk6d.cloudfr... [cloudfront.net]
Complete outsider... (Score:2)
Re: (Score:2)
I believe you can use it for free too... I don't really know what the advantages of giving them money, or the disadvantages of not doing so are.
Re: Complete outsider... (Score:2)
Re: (Score:2)
Free tier is limited to ~10-20 queries per day of the low end models, whereas $20 tier is virtually unlimited for the "dumbest" models , and moderate access to reasoning models, while $200 tier gives you access to enhanced reasoning models that can think for multiple minutes at a time
Re: (Score:1)
Free tier is limited to ~10-20 queries per day of the low end models, whereas $20 tier is virtually unlimited for the "dumbest" models , and moderate access to reasoning models, while $200 tier gives you access to enhanced reasoning models that can think for multiple minutes at a time
That's a reasonable summary but I think it downplays the $20/month access (Plus subscription) a little too much. For $20/month you get quota-based access to the best models (including the "Deep Research" ones that spend 30mins developing a research paper). My understanding is that for $200/month (Pro subscription) you get unlimited access to the best models.
Re: (Score:2)
I pay $20/mo for chat access to an llm and $20/mo for a private search engine. $40/mo to escape the googleverse is well worth the price in my opinion.
Re: Complete outsider... (Score:2)
Re: (Score:2)
Depends on what kind of AI and what you're doing with it. ChatGPT costs about 20 bucks and has a free version. LLM-APIs cost between $0.01 and $20 per million tokens depending on what model you need. It's harder to tell for image AI as most seems to be integrated with other products.
Percentages are a key sign of bullshit (Score:3, Funny)
I'm the leader of the fastest growing religion with over 400% growth over the past two years! I mean, we went from one to four members over that time, but nobody else is growing faster!
Re: Percentages are a key sign of bullshit (Score:2)
do any of your three new members understand percentages?
Re: (Score:2)
Doubtful, but they are excited enough to give a 133% effort.
Not surprising (Score:2)
Exec's are pretty high on the possibility of firing everyone and are pushing us to use it.
Re: (Score:2)
Exec's are pretty high on the possibility of firing everyone and are pushing us to use it.
Aww, I was hoping to replace myself with a very small shell script then still collect checks.
Not even once quarterly. (Score:2)
A few times annually is once every four months. TFA does have stats for "a few times a week" users as well. If AI was really all that, lots of people would be using it dozens of times a day. But they didn't study that because reasons. Something that helps a few times a week doesn't sounds like something that can reduce staffing and costs more than a few percent.
Re: (Score:2)
The headline, "AI use at work has nearly doubled", is literally only applying to the daily stat they collected, which increased from 4% to 8%.
The "a few times a week" and "a few times a year" stats didn't double (though increased)
I'm really struggling to see how you missed that.
AI workplace (Score:2)
2 years ago, ChatGPT was just 6 months old (Score:3, Insightful)
So it kind of makes sense that usage is up by 100% since then. Two years ago, only nerds like those on Slashdot, had even heard of ChatGPT. Now, everybody is using it.
I think that number is low, but whatever.
And it ain't good (Score:3)
I damn near took a fit when I read it. The focus was wrong, congratulating someone not even involved with the project, had several things in it that were just plain wrong, and assigned several quotes to me that I never said.
Then they had the audacity to claim that my quotes were okay since the AI made them positive.
It's taking a bit of time to get used to the AI driven post truth era. We're like the guy who drove around several blockades then drove off a bridge because his GPS told him to.