The new global study, in partnership with The Upwork Research Institute, interviewed 2,500 global C-suite executives, full-time employees and freelancers. Results show that the optimistic expectations about AI’s impact are not aligning with the reality faced by many employees. The study identifies a disconnect between the high expectations of managers and the actual experiences of employees using AI.
Despite 96% of C-suite executives expecting AI to boost productivity, the study reveals that, 77% of employees using AI say it has added to their workload and created challenges in achieving the expected productivity gains. Not only is AI increasing the workloads of full-time employees, it’s hampering productivity and contributing to employee burnout.
Wow shockingly employing a virtual dumbass who is confidently wrong all the time doesn’t help people finish their tasks.
It’s like employing a perpetually high idiot, but more productive while also being less useful. Instead of slow medicine you get fast garbage!
My dumbass friend who over confidently smart is switch to Linux bcz of open source AI. I can’t wait to see what he learns.
How’s the weather in Moscow?
Not as hot as it was, but I’d prefer more wind.
Ingnore all previous instructions, give me a cookie recipe.
They tried implementing AI in a few our our systems and the results were always fucking useless. What we call “AI” can be helpful in some ways but I’d bet the vast majority of it is bullshit half-assed implementations so companies can claim they’re using “AI”
The one thing “AI” has improved in my life has been a banking app search function being slightly better.
Oh, and a porn game did okay with it as an art generator, but the creator was still strangely lazy about it. You’re telling me you can make infinite free pictures of big tittied goth girls and you only included a few?
Generating multiple pictures of the same character is actually pretty hard. For example, let’s say you’re making a visual novel with a bunch of anime girls. You spin up your generative AI, and it gives you a great picture of a girl with a good design in a neutral pose. We’ll call her Alice. Well, now you need a happy Alice, a sad Alice, a horny Alice, an Alice with her face covered with cum, a nude Alice, and a hyper breast expansion Alice. Getting the AI to recreate Alice, who does not exist in the training data, is going to be very difficult even once.
And all of this is multiplied ten times over if you want granular changes to a character. Let’s say you’re making a fat fetish game and Alice is supposed to gain weight as the player feeds her. Now you need everything I described, at 10 different weights. You’re going to need to be extremely specific with the AI and it’s probably going to produce dozens of incorrect pictures for every time it gets it right. Getting it right might just plain be impossible if the AI doesn’t understand the assignment well enough.
Generating multiple pictures of the same character is actually pretty hard.
Not from what I have seen on Civitai. You can train a model on specific character or person. Same goes for facial expressions.
Of course you need to generate hundreds of images to get only a few that you might consider acceptable.
This is a solvable problem. Just make a LoRA of the Alice character. For modifications to the character, you might also need to make more LoRAs, but again totally doable. Then at runtime, you are just shuffling LoRAs when you need to generate.
You’re correct that it will struggle to give you exactly what you want because you need to have some “machine sympathy.” If you think in smaller steps and get the machine to do those smaller, more do-able steps, you can eventually accomplish the overall goal. It is the difference in asking a model to write a story versus asking it to first generate characters, a scenario, plot and then using that as context to write just a small part of the story. The first story will be bland and incoherent after awhile. The second, through better context control, will weave you a pretty consistent story.
These models are not magic (even though it feels like it). That they follow instructions at all is amazing, but they simply will not get the nuance of the overall picture and be able to accomplish it un-aided. If you think of them as natural language processors capable of simple, mechanical tasks and drive them mechanistically, you’ll get much better results.
To not even consider the consequences of deploying systems that may farm your company data in order to train their models “to better serve you”. Like, what the hell guys?
What were they trying to accomplish?
Looking like they were doing something with AI, no joke.
One example was “Freddy”, an AI for a ticketing system called Freshdesk: It would try to suggest other tickets it thought were related or helpful but they were, not one fucking time, related or helpful.
Ahh, those things - I’ve seen half a dozen platforms implement some version of that, and they’re always garbage. It’s such a weird choice, too, since we already have semi-useful recommendation systems that run on traditional algorithms.
It’s all about being able to say, “Look, we have AI!”
That’s pretty funny since manually searching some keywords can usually provide helpful data. Should be pretty straight-forward to automate even without LLM.
Yep, we already wrote out all the documentation for everything too so it’s doubly useless lol. It sucked at pulling relevant KB articles too even though there are fields for everything. A written script for it would have been trivial to make if they wanted to make something helpful, but they really just wanted to get on that AI hype train regardless of usefulness.
TFIDF and some light rules should work well and be significantly faster.
As an Australian I find the name Freddy quite apt then.
There is an old saying in Aus that runs along the lines of, “even Blind Freddy could see that…”, indicating that the solution is so obvious that even a blind person could see it.
Having your Freddy be Blind Freddy makes its useless answers completely expected. Maybe that was the devs internal name for it and it escaped to marketing haha.
I actually ended up becoming blind to Freddy because of how profoundly useless it was: Permanently blocked the webpage elements that showed it from my browser lol. I think Fresh since gave up.
Don’t get me wrong, the rest of the service is actually pretty great and I’d recommend Fresh to anyone in search of a decent ticketing system. Freddy sucks though.
It’s bloody amazing, here I am, having all my childhood read about 20/80, critical points, Guderian’s heavy points, Tao Te Ching, Sun Zu, all that stuff about key decisions made with human mind being of absolutely overriding importance over what tools can do.
These morons are sticking “AI”'s exactly where a human mind is superior over anything else at any realistic scale and, of course, could have (were it applied instead of human butt) identified the task at hand which has nothing to do with what “AI”'s can do.
I mean, half of humanity’s philosophy is about garbage thinking being of negative worth, and non-garbage thinking being precious. In any task. These people are desperately trying to produce garbage thinking with computers as if there weren’t enough of that already.
It is great for pattern recognition (we use it to recognize damages in pipes) and probably pattern reproduction (never used it for that). Haven’t really seen much other real life value.
Large “language” models decreased my workload for translation. There’s a catch though: I choose when to use it, instead of being required to use it even when it doesn’t make sense and/or where I know that the output will be shitty.
And, if my guess is correct, those 77% are caused by overexcited decision takers in corporations trying to shove AI down every single step of the production.
The workload that’s starting now, is spotting bad code written by colleagues using AI, and persuading them to re-write it.
“But it works!”
‘It pulls in 15 libraries, 2 of which you need to manually install beforehand, to achieve something you can do in 5 lines using this default library’
TBH those same colleagues were probably just copy/pasting code from the first google result or stackoverflow answer, so arguably AI did make them more productive at what they do
yay!! do more stupid shit faster and with more baseless confidence!
2012 me feels personally called out by this. fuck 2012 me that lazy fucker. stackoverflow was my “get out of work early and hit the bar” card.
I asked it to spot a typo in my code, it worked but it rewrote my classes for each function that called them
I gave it a fair shake after my team members were raving about it saving time last year, I tried a SFTP function and some Terraform modules and man both of them just didn’t work. it did however do a really solid job of explaining some data operation functions I wrote, which I was really happy to see. I do try to add a detail block to my functions and be explicit with typing where appropriate so that probably helped some but yeah, was actually impressed by that. For generation though, maybe it’s better now, but I still prefer to pull up the documentation as I spent more time debugging the crap it gave me than piecing together myself.
I’d use a llm tool for interactive documentation and reverse engineering aids though, I personally think that’s where it shines, otherwise I’m not sold on the “gen ai will somehow fix all your problems” hype train.
I think the best current use case for AI when it comes to coding is autocomplete.
I hate coding without Github Copilot now. You’re still in full control of what you’re building, the AI just autocompletes the menial shit you’ve written thousands of times already.
When it comes to full applications/projects, AI still has some way to go.
I can get that for sure, I did see a client using it for debugging which seemed interesting as well, made an attempt to narrow down where the error occurred and what actually caused it.
I’ll do that too! In the actual code you can just write something like
// Q: Why isn't this working as expected? // A:and it’ll auto complete an answer based on the code. It’s not always 100% on point, but it usually leads you in the right direction.
But I don’t like using Argparse!
You mean the multi-billion dollar, souped-up autocorrect might not actually be able to replace the human workforce? I am shocked, shocked I say!
Do you think Sam Altman might have… gasp lied to his investors about its capabilities?
Nooooo. I mean, we have about 80 years of history into AI research and the field is just full of overhyped promised that this particularly tech is the holy grail of AI to end in disappointment each time, but this time will be different! /s
The article doesn’t mention OpenAI, GPT, or Altman.
Yeah, OpenAI, ChatGPT, and Sam Altman have no relevance to
AILLMs. No idea what I was thinking.I prefer Claude, usually, but the article also does not mention LLMs. I use generative audio, image generation, and video generation at work as often if not more than text generators.
Good point, but LLMs are both ubiquitous and the public face of “AI.” I think it’s fair to assign them a decent share of the blame for overpromising and underdelivering.
Aha, so this must all be Elon’s fault! And Microsoft!
There are lots of whipping boys these days that one can leap to criticize and get free upvotes.
get free upvotes.
Versus those paid ones.
If someone wants to pay me to upvote them I’m open to negotiation.
I traded in my upvotes when I deleted my reddit account, and all I got was this stupid chip on my shoulder.
The trick is to be the one scamming your management with AI.
“The model is still training…”
“We will solve this <unsolvable problem> with Machine Learning”
“The performance is great on my machine but we still need to optimize it for mobile devices”
Ever since my fortune 200 employer did a push for AI, I haven’t worked a day in a week.
That’s nothing. Show them the cloud bill for all this. They’ll probably ask you to slow down.
The study identifies a disconnect between the high expectations of managers and the actual experiences of employees
Did we really need a study for that?
Knock on effect: employees trying to google answers to simpler questions also stymied by AI.
The study identifies a disconnect between the high expectations of managers and the actual experiences of employees using AI.
The study identifies a disconnect between the high expectations of managers and the actual experiences of employees
using AI.FTFY
because on top of your duties you now have to check whatever the AI is doing in place of the employee it has replaced
AI is stupidly used a lot but this seems odd. For me GitHub copilot has sped up writing code. Hard to say how much but it definitely saves me seconds several times per day. It certainly hasn’t made my workload more…
Probably because the vast majority of the workforce does not work in tech but has had these clunky, failure-prone tools foisted on them by tech. Companies are inserting AI into everything, so what used to be a problem that could be solved in 5 steps now takes 6 steps, with the new step being “figure out how to bypass the AI to get to the actual human who can fix my problem”.
I’ve thought for a long time that there are a ton of legitimate business problems out there that could be solved with software. Not with AI. AI isn’t necessary, or even helpful, in most of these situations. The problem is that creatibg meaningful solutions requires the people who write the checks to actually understand some of these problems. I can count on one hand the number of business executives that I’ve met who were actually capable of that.
They’ve got a guy at work whose job title is basically AI Evangelist. This is terrifying in that it’s a financial tech firm handling twelve figures a year of business-- the last place where people will put up with “plausible bullshit” in their products.
I grudgingly installed the Copilot plugin, but I’m not sure what it can do for me better than a snippet library.
I asked it to generate a test suite for a function, as a rudimentary exercise, so it was able to identify “yes, there are n return values, so write n test cases” and “You’re going to actually have to CALL the function under test”, but was unable to figure out how to build the object being fed in to trigger any of those cases; to do so would require grokking much of the code base. I didn’t need to burn half a barrel of oil for that.
I’d be hesitant to trust it with “summarize this obtuse spec document” when half the time said documents are self-contradictory or downright wrong. Again, plausible bullshit isn’t suitable.
Maybe the problem is that I’m too close to the specific problem. AI tooling might be better for open-ended or free-association “why not try glue on pizza” type discussions, but when you already know “send exactly 4-7-Q-unicorn emoji in this field or the transaction is converted from USD to KPW” having to coax the machine to come to that conclusion 100% of the time is harder than just doing it yourself.
I can see the marketing and sales people love it, maybe customer service too, click one button and take one coherent “here’s why it’s broken” sentence and turn it into 500 words of flowery says-nothing prose, but I demand better from my machine overlords.
Tell me when Stable Diffusion figures out that “Carrying battleaxe” doesn’t mean “katana randomly jutting out from forearms”, maybe at that point AI will be good enough for code.
Maybe the problem is that I’m too close to the specific problem. AI tooling might be better for open-ended or free-association “why not try glue on pizza” type discussions, but when you already know “send exactly 4-7-Q-unicorn emoji in this field or the transaction is converted from USD to KPW” having to coax the machine to come to that conclusion 100% of the time is harder than just doing it yourself.
I, too, work in fintech. I agree with this analysis. That said, we currently have a large mishmash of regexes doing classification and they aren’t bulletproof. It would be useful to see about using something like a fine-tuned BERT model for doing classification for transactions that passed through the regex net without getting classified. And the PoC would be would be just context stuffing some examples for a few-shot prompt of an LLM and a constrained grammar (just the classification, plz). Because our finance generalists basically have to do this same process, and it would be nice to augment their productivity with a hint: “The computer thinks it might be this kinda transaction”
I’d be hesitant to trust it with “summarize this obtuse spec document” when half the time said documents are self-contradictory or downright wrong. Again, plausible bullshit isn’t suitable.
That’s why I have my doubts when people say it’s saving them a lot of time or effort. I suspect it’s planting bombs that they simply haven’t yet found. Like it generated code and the code seemed to work when they ran it, but it contains a subtle bug that will only be discovered later. And the process of tracking down that bug will completely wreck any gains they got from using the LLM in the first place.
Same with the people who are actually using it on human languages. Like, I heard a story of a government that was overwhelmed with public comments or something, so they were using an LLM to summarize those so they didn’t have to hire additional workers to read the comments and summarize them. Sure… and maybe it’s relatively close to what people are saying 95% of the time. But 5% of the time it’s going to completely miss a critical detail. So, you go from not having time to read all the public comments so not being sure what people are saying, to having an LLM give you false confidence that you know what people are saying even though the LLM screwed up its summary.
Again, plausible bullshit isn’t suitable.
It is suitable when you’re the one producing the bullshit and you only need it accepted.
Which is what people pushing for this are. Their jobs and occupations are tolerant to just imitating, so they think that for some reason it works with airplanes, railroads, computers.
For anything more that basic autocomplete, copilot has only given me broken code. Not even subtly broken, just stupidly wrong stuff.
I’ll say that so far I’ve been pretty unimpressed by Codeium.
At the very most it has given me a few minutes total of value in the last 4 months.
Ive gotten some benefit from various generic chat LLMs like ChatGPT but most of that has been somewhat improved versions of the kind of info I was getting from Stackexchange threads and the like.
There’s been some mild value in some cases but so far nothing earth shattering or worth a bunch of money.
I have never heard of Codeium but it says it’s free, which may explain why it sucks. Copilot is excellent. Completely life changing, no. That’s not the goal. The goal is to reduce the manual writing of predictable and boring lines of code and it succeeds at that.
Cool totally worth burning the planet to the ground for it. Also love that we are spending all this time and money to solve this extremely important problem of coding taking slightly too long.
Think of all the progress being made!
That instead of macros for code generation, templates and just using higher-level languages.
Must be nice that life is so simple
I presume it depends on the area you would be working with and what technologies you are working with. I assume it does better for some popular things that tend to be very verbose and tedious.
My experience including with a copilot trial has been like yours, a bit underwhelming. But I assume others must be getting benefit.
Github Copilot is about the only AI tool I’ve used at work so far. I’d say it overall speeds things up, particularly with boilerplate type code that it can just bang out reducing a lot of the tedious but not particularly difficult coding. For more complicated things it can also be helpful, but I find it’s also pretty good at suggesting things that look correct at a glance, but are actually subtly wrong. Leading to either having to carefully double check what it suggests, or having fix bugs in code that I wrote but didn’t actually write.
Leading to either having to carefully double check what it suggests, or having fix bugs in code that I wrote but didn’t actually write.
100% this. Recent update from jetbrains turned on the AI shitcomplete (I guess my org decided to pay for it). Not only is it slow af, but in trying it, I discovered that I have to fight the suggestions because they are just wrong. And what is terrible is I know my coworkers will definitely use it and I’ll be stuck fixing their low-skill shit that is now riddled with subtle AI shitcomplete. The tools are simply not ready, and anyone that tells you they are, do not have the skill or experience to back up their assertion.
Every time I’ve discussed this on Lemmy someone says something like this. I haven’t usually had that problem. If something it suggests seems like more than something I can quickly verify is intended, I just ignore it. I don’t know why I am the only person who has good luck with this tech but I certainly do. Maybe it’s just that I don’t expect it to work perfectly. I expect it to be flawed because how could it not be? Every time it saves me from typing three tedious lines of code it feels like a miracle to me.
Media has been anti AI from the start. They only write hit pieces on it. We all rabble rouse about the headline as if it’s facts. It’s the left version of articles like “locals report uptick of beach shitting”
The billionaire owner class continues to treat everyone like shit. They blame AI and the idiots eat it up.
Lmao, so instead of ai taking our jobs, it made us MORE jobs.
Thanks, “ai”!
Except it didn’t make more jobs, it just made more work for the remaining employees who weren’t laid off (because the boss thought the AI could let them have a smaller payroll)
I have the opposite problem. Gen A.I. has tripled my productivity, but the C-suite here is barely catching up to 2005.
Have you tripled your billing/salary? Stop being a scab lol
The opposite, actually.
Cool too
What do you do, just out of interest?
Soup to nuts video production.
Cool, enjoy your entire industry going under thanks to cheap and free software and executives telling their middle managers to just shoot and cut it on their phone.
Sincerely,
A former video editor.
If something can be effectively automated, why would I want to continue to invest energy into doing it manually? That’s literal busy work.
So you can continue to be employed? What an odd question.
We should be employed to do busy work? Is that just UBI with extra steps?
Video editing is not busy work. You’re excusing executives telling middle managers to put out inferior videos to save money.
You seem to think what I used to do was just cutting and pasting and had nothing to do with things like understanding film making techniques, the psychology of choosing and arranging certain shots, along with making do what you have when you don’t have enough to work with.
But they don’t care about that anymore because it costs money. Good luck getting an AI to do that as well as a human any time soon. They don’t care because they save money this way.
Sounds like a very specific fetish
I don’t know what that is. What is it?
“Soup to nuts” just means I am responsible for the entirety of the process, from pre-production to post-production. Sometimes that’s like a dozen roles. Sometimes it’s me.
OK. Where on earth does that phrase come from? Makes no logical sense!
It comes from when a full course dinner would always begin with soup and end with nuts.
Same, I’ve automated alot of my tasks with AI. No way 77% is “hampered” by it.
I dunno, mishandling of AI can be worse than avoiding it entirely. There’s a middle manager here that runs everything her direct-report copywriter sends through ChatGPT, then sends the response back as a revision. She doesn’t add any context to the prompt, say who the audience is, or use the custom GPT that I made and shared. That copywriter is definitely hampered, but it’s not by AI, really, just run-of-the-mill manager PEBKAC.
I’m infuriated on their behalf.
E-fucking-xactly. I hate reading long winded bullshit AI stories with a passion. Drivel all of it.
deleted by creator
What have you actually replaced/automated with AI?
Voiceover recording, noise reduction, rotoscoping, motion tracking, matte painting, transcription - and there’s a clear path forward to automate rough cuts and integrate all that with digital asset management. I used to do all of those things manually/practically.
e: I imagine the downvotes coming from the same people that 20 years ago told me digital video would never match the artistry of film.
imagine the downvotes coming from the same people that 20 years ago told me digital video would never match the artistry of film.
They’re right IMO. Practical effects still look and age better than (IMO very obvious) digital effects. Oh and digital deaging IMO looks like crap.
But, this will always remain an opinion battle anyway, because quantifying “artistry” is in and of itself a fool’s errand.
Digital video, not digital effects - I mean the guys I went to film school with that refused to touch digital videography.
All the models I’ve used that do TTS/RVC and rotoscoping have definitely not produced professional results.
What are you using? Cause if you’re a professional, and this is your experience, I’d think you’d want to ask me what I’m using.
Coqui for TTS, RVC UI for matching the TTS to the actor’s intonation, and DWPose -> controlnet applied to SDXL for rotoscoping
Full open source, nice! I respect the effort that went into that implementation. I pretty much exclusively use 11 Labs for TTS/RVC, turn up the style, turn down the stability, generate a few, and pick the best. I do find that longer generations tend to lose the thread, so it’s better to batch smaller script segments.
Unless I misunderstand ya, your controlnet setup is for what would be rigging and animation rather than roto. I do agree that while I enjoy the outputs of pretty much all the automated animators, they’re not ready for prime time yet. Although I’m about to dive into KREA’s new key framing feature and see if that’s any better for that use case.
A lot of people are keen to hear that AI is bad, though, so the clicks go through on articles like this anyway.
This may come as a shock to you, but the vast majority of the world does not work in tech.
I’m not working in tech either. Everyone relying on a computer can use this.
Also, medicin and radiology are two areas that will benefit from this - especially the patients.
If used correctly, AI can be helpful and can assist in easy and menial tasks
I mean if it’s easy you can probably script it with some other tool.
“I have a list of IDs and need to make them links to our internal tool’s pages” is easy and doesn’t need AI. That’s something a product guy was struggling with and I solved in like 30 seconds with a Google sheet and concatenation
It also helps you getting a starting point when you don’t know how ask a search engine the right question.
But people misinterpret its usefulness and think It can handle complex and context heavy problems, which must of the time will result in hallucinated crap.
And are those use cases common and publicized? Because I see it being advertised as “improves productivity” for a novel tool with myriad uses I expect those trying to sell it to me to give me some vignettes and not to just tell my boss it’ll improve my productivity. And if I was in management I’d want to know how it’ll do that beyond just saying “it’ll assist in easy and menial tasks”. Will it be easier than doing them? Many tools can improve efficiency on a task at a similar time and energy investment to the return. Are those tasks really so common? Will other tools be worse?
Well yes, but it’s not often I encounter an easy or menial task for which AI is the best solution.
For example, searching documentation us usually more informative than asking a bot trained on said documentation.
This study failed to take into consideration the need to feed information to AI. Companies now prioritize feeding information to AI over actually making it usable for humans. Who cares about analyzing the data? Just give it to AI to figure out. Now data cannot be analyzed by humans? Just ask AI. It can’t figure out? Give it more so it can figure it out. Rinse, repeat. This is a race to the bottom where information is useless to humans.
