Their primary use case in the office that I’ve seen is asking someone a question and having them send a LLM response where they clearly didn’t read what you asked and the response they sent you does not answer the original question. It’s so cool!
Comment on Microsoft ends OpenAI exclusivity in Office, adds rival Anthropic
gedaliyah@lemmy.world 10 hours ago
What are people actually using AI for in office programs? I spend almost all day working on email, documents, spreadsheets, and presentations. Every time in the past that I have tried to use AI tools, it has either failed to perform the task, done it poorly/incorrectly, or taken longer than using traditional tools (requiring multiple prompts, editing and correcting, etc.)
Am I just a dinosaur? Are there people who really use these tools productively?
GnuLinuxDude@lemmy.ml 8 hours ago
gedaliyah@lemmy.world 7 hours ago
That has sorta been my experience so far. LLMs are great at producing output as long as the quality of the output doesn’t really matter. Maybe there are a lot more tasks than I realize where this is the case - in my work there are not many.
CompostMaterial@lemmy.world 9 hours ago
Document summary, quick email replies that sound coherent and professional when you have one foot out the door on a Friday, if the Org has fabric then the PowerPoint copilot can be a real time save of given me a presentation based off this outline and use some data as the source.
I find that those who are office professionals that don’t like AI most often either have expectations of perfection or don’t know how to prompt properly.
Ai excels at get you 60-70% of the way there the first time. Then you can either refine it with additional prompts or manually. Either way if it saves me time and effort on something I care nothing about then it is worth it to me.
homesweethomeMrL@lemmy.world 9 hours ago
Depending on the ask, yes. Summarize this document - usually pretty correct, if verbose and disgustingly bland like eating paste. But then it’ll say “Do you want me to make this into a powerpoint?” and you go - sure, why not - and (true story) it goes “here it is” and there’s nothing there.
And you say “uh, there’s nothing there” and it goes “oh! Sorry. Here it is” and it links to a pdf. And you say “that’s not a powerpoint” and it goes “oh! sorry you’re right. Here.” and it links to a powerpoint that’s so sparse it seems like it assembled itself from the random background radiation of the universe - and it doesn’t even say what the summary said. It’s something sort of like it but worse somehow.
And that experience never changes. That’s essentially all any AI interaction will ever be.
GnuLinuxDude@lemmy.ml 8 hours ago
The secondary use here is when you get asked in performance evaluation time how much you’re incorporating AI into your workflow. And you’d better not say “none” so you keep generating these nonsense documents and throwing them away so it appears like you’re using AI
homesweethomeMrL@lemmy.world 8 hours ago
Yeah. “How Every Business Is Screwing Itself By Requiring What It Doesn’t Understand”
CompostMaterial@lemmy.world 8 hours ago
I agree that that is an experience I have frustratingly had as well. I disagree that it won’t change. Agentic Ai allows for specialized services to be integrated. All it would take is an agent to be created that specializes in creating pptx documents and the integrated into the primary service. The capability already exists it just need to be implemented correctly.
homesweethomeMrL@lemmy.world 8 hours ago
Given the unprecedented amount of sheer capital rolling into this like an ocean falling down from space, over what, three years now? A subroutine to “make pptx” not existing doesn’t give me a ton of hope that anyone either knows what they’re doing or is able and willing to chop enough of it off to do that.
I hear it in concept, I’m just saying - it seems like such low hanging fruit it’s almost damning that it doesn’t exist.
homesweethomeMrL@lemmy.world 9 hours ago
I use them for cursory web searches it would take me slightly longer to do directly.
spellings of words, name of the character from that movie, whatever ephemera I want at the moment that no one ever should really give two shits about.
Actual questions or actual functions? lol no. And the second they ask me to pay for it - it’s gone.
jaybone@lemmy.zip 8 hours ago
Yeah the only thing I find useful is the web search result when I want like a 5 line code example.
Problem is they could do the same thing by just scraping and parsing the first result and inlining it in the search result page so I don’t have to click on a link. Same thing just without wasting all that hardware and energy.
homesweethomeMrL@lemmy.world 7 hours ago
Yep. It’s not even very good at the barest minimum function anyone can find for it after years of blasting everyone with propaganda to use it. And could be replaced with a couple of lines.
Truly a bizarre chapter in technology.
gedaliyah@lemmy.world 7 hours ago
I can see a use case when it comes to search - like you said. If the question is relatively trivial and slightly obscure then a LLM summary is probably adequate (maybe problematic for other reasons).
But this is being marketed in Productivity software! I really and genuinely want to understand why it seems so popular.
homesweethomeMrL@lemmy.world 7 hours ago
So far as I can tell there are a few groups that make up its “popularity”
Ironically the “anti-AI” crowd, for lack of a more at-hand description mention AI a lot because it’s a form of hype many experienced technology-versed people have seen many times before and at such a scale as to be astounding. Also because it doesn’t seem to work very well at all. And because everyone is being forced to use it and “forced” to like it.
Then there’s the gung-ho “newbies” (who may in fact have more than a decade of experience) who just think it’s neat. They love the intricacy of it and the wide-open futuristic vistas and they like talking about what it can or might be able to do some day. This group will become the former group in another decade or so.
Then there’s the propaganda - any news article, tweet, comment or what have you that in some way shape or form is being put in front of many people for the purposes of keeping AI discussion happening. Much of this is “free press” because next to lawyers, journalists are least likely to understand technology, and their directives from on high are to promote this thing - whatever it is.
Then there’s data science researchers who are ostensibly pulling new insights out of a sea of hallucinations and making interesting new connections because of it. It happens.
I think that’s more or less it. There’s also the general non-technology group of people who are told things about it and have not used it, nor will they probably ever.
Imgonnatrythis@sh.itjust.works 9 hours ago
They push it so hard and yet I agree, its real world utility is incredibly low.
jj4211@lemmy.world 9 hours ago
Feel like I’m being gaslit, the more I try to use them the less confident I become in their utility.
I will confess it did help me cut through some particularly obtuse documentation to provide a rough example of what I wanted to do. It still totally screwed up the actual suggestion, but it at least helped me figure out some good keywords to dig into.
It occasionally saves me some tedium when I have to do something mindlessly tedious, but doing that usually also inflicts constant misguesses about what next.
But even when doing easy stuff they are falling over constantly, and that hasn’t been significantly improving.