I’ve noticed a correlation between increased workloads and employee layoffs. Essentially, companies seem to be replacing 50% of their workforce with AI and expecting the remaining employees to take on the additional responsibilities of those who were let go.
Tbf that's exactly what productivity means - with technology one farmer can now do work that once took hundreds of labourers. Problem is that right now the landowners have decided their new magic ploughs and sickles are twice as good so half the peasants can be used as target practice and the rest are expected to bring in the harvest with half as many hands. Workload obviously goes up when the masters demand more work from fewer people on less pay. And they'll seize any reason to demand it.
This is a problem with not letting companies fail when they make poor decisions. We need these companies to fail so that these middle managers and executives don't keep piling up and adding parasitic undergrowth. And then you let new companies take their spot that actually make good decisions.
Fire 50% of their work force, then wonder why the remaining employees are complaining they're being worked to the bone. I never worried so much about losing my job to AI, I always worried management would listen to some 20 minute lecture and be fooled into downsizing, thinking it was the solution to all their staffing problems. Imagine my surprise, seeing it unfold in real time.
@@Cye_Pieit does work though, these managers have golden parachutes to get into another company and they know it, all the fallback and consequences go to the junior staff/ long term employess outside of management, they can afford to not care with connections, it's not that they are incompetant, they know they can get away with it and repeat it somewhere else
My mom is a tax professional working at a multinational corporation. The job itself is stressful enough. She was in a department of 5 where they had this fresh grad just doing data entry. The kid was great, I met him a few times. Just sits there quietly, doing his work. 99.5% accurately from what I've been told. Some AI startup came along, promised to eliminate his position and many others, management jumped on it and the AI is absolute dog doo doo. It's only about 10% accurate and costs more than the kid alone lol. The corporation has decided to stop using the AI but refused to hire on a person which makes zero sense. My mom who hasn't done data entry in decades is now stuck doing Jr level work, while also being expected to complete her own duties. This **** is ridiculous. She is about a molecule away form walking out and I can't wayt for that day. They'll have to hire at least 2 people at her salary each to get the same amount of work done. I saw job postings from them but didn't bother applying. F em.
Yeah AI doesn't belong in any area requiring 100% accuracy. This is a well-known limitation. It is great when something that looks right and is mostly correct is what is needed. You'd expect an AI startup or your company management to know this. It isn't a secret or anything requiring deep knowledge.
As someone who has worked at several tech companies, I can tell you those years were unbelievably unproductive. I was worked into oblivion, don’t get me wrong, but I was working on a whole lot of nothing. I was expected to learn countless pieces of software that were supposed to help us stay organized, make us more efficient, and that kind of thing. They never did. Just made work more difficult. No one ever followed the process. Everyone pretty much just made their own rules. I remember I was using a project management software at one company and at one point another department decided to use a different project management software. We then had to learn their project management software because we always worked with them. Instead of building out projects in one piece of software, now we had to build it in two. I had to check off tasks in two softwares and oh by the way, had to send an email and a slack message on top of that to our stakeholders anytime a project was finished, then I’d have to monitor all 4 channels to capture any feedback. It was just absolutely ridiculous.
Ohhhhh, feeling like that now. Between Microsoft teams, email, airtable and a "custom software" for our company, I, the human, am ensuring the same information existing in like 5 formats for different people and me making sure they all stay the same along the way with this software that was suppose to streamline the process, but doesn't.
@@T1Oracle But why, when we can chase buzzwords for guaranteed success and have fingers in our ears to all those boring complainers? Like don't they realize they are killing our mood with all their reasonable concerns?
My org is trying to implement AI to replace humans, upper-management has blocked off negative feedback about AI so that they can continue overpromising so that C-Suite doesn't hear anything except that it's cutting costs. We're all hauling ass working overtime to try and keep up with the promised speed of AI. It's like the car in flintstones where it's really just a few dudes pedaling for their life.
@@Artifex_Prudens sure you can argue that but they used to be paid MUCH less(orders of magnitude) and were doing just fine. In comparison a lot of the working force is one step away from being homeless
There is one major reason why freelancers are more efficient than full-time employees, freelancers don't have useless, and knowledgeable management and higher-ups implementing wasteful, redundant, and pointless policies onto their employees. I worked a retail job where my certain department was so small, just me and another guy, that they removed the Manager position for our department. Everything immediately improved, because the people responsible for making the big decisions where the people doing the actual job
I moved from an amazing workplace where our manager was maybe 40% manager and 60% on the floor working, to a workplace where the manager (same job) was only manager and had no idea what we were doing. It was so bad. He didn't know about safety, how long things took, what we were even doing or talking about. Everyone were complaining and eventually I quit.
Tech has a massive, massive, systemic pattern of just throwing the newest thing at every problem when it wasn't designed to solve every problem and in fact can make many problems harder to solve, not easier. It's very odd to me that with so many so-called "smart" and "genius" people that we have, we all collectively do this over and over and over and expect different results.
See it's less that tech people are actually smarter and more that "being smart" is intrinsically tied to their sense of self worth, so they will often pile onto the thing that makes them seen as smart 😪 many are of course smart, but not too much smarter than law or finance or medicine people
i mean, if they receive millions in venture capital funding to use a hammer to drill screws, then i would say the venture capitalists are the idiots, not the startups
It's funny that managers and companies will complain about employees not coming back to face to face work but then hire a freelancer whom they never even see and could be in an entirely different country!!! WTF!!!
Because it's usually about 10x cheaper. So they can hire 3-5 people instead of one locally and save half their salary, even though those 3-5 people will do shitty work.
When Autocad first started it was considered revolutionary in respect to drawings and drafting. It was expected to reduce the workload for draftsman as replication and copying was greatly simplified. As exactly mentioned, expectations on what can be produced sky-rocketed at the same time.
About freelancers and productivity: as a freelancer you can afford to not give crap about quality. You will not have to maintain that code for a long time. And when that code explodes or gets expoited, you are long gone.
Yes, this means you can be negatively productive where your code requires more time to fix than it took to write it. As a field, we don't need more code. We need code to be correct. We need code to be simplified. But we really don't need more KLOCs.
@@username7763its not a freelancer faults tho. We get hired for specific task not knowing or having time to dig deep into the software. And free market is cheaper and faster the better. And then we tend to not come back to same project because next person is now cheaper than we. Full time is always better than part time. Part time is good for temporary support but needs to lead and managed by ppl working on a project. We just lost our 4th freelance dev we relied upon. That can hit you back too. But then again we never had enough work to sustain him full time or even part time. Company we deal with lost their full time and instead of hiring new full time they rely on freelance to do the same that this person been doing full time for years.
that is called a scam artist. Where I'm from once you develop it you support it on a SLA retainer. Any free lancer that develops and disappears is a >on, you develop with the aim to get more SLA and recurring income.
On the point of farmers working 6-7 day workweeks, they also typically stopped working when the sun went down and during winter (in colder climates) not to mention not everyone was a farmer at the time
Work for 4 hours, finish up, do some house chores, is not the same as a warehouse telling people to stand around and push boxes for 12 and thrn go home, make their own food, chores, etc
Also, a lot of time on the farm is waiting. Waiting for the time to do something. The crops need time to grow. Farmer can only do so much. There are other tasks but not so much
The CEO of one of the biggest Nordic traveI agencies canceIIed a very much needed e-commerce website and database upgrade project and diverted aII resources to instead focus on "APIs" because he had heard about something caIIed "APIs" on a conference. That project faiIed. Many many miIIions down the drain. Situation got so fubar that the shareholders had to seII the company roughly a year later, and thus have the CEO fired. Ironically, he was repIaced by the previous CEO of the company, that's how bad he was.
@@Daniel_WR_Hart That's one of the many questions that we (architects and developers) relayed to middle management, whom in turn completely ignored those questions in discussions with the C-level managers.
Honestly I still don't get what "AI-native" means. It probably isn't hard to figure out, my brain just has a "yuck" reaction to that word combination and refuses to get it.
@@alexandrecolautoneto7374 Yeah, probably. I just don't like to dismiss things I don't understand as nonsense before I look into them, and I'm not exactly motivated to look into "AI-native".
> who imagined that the only developer left would take the work of the 5 fired for AI-native? The real question is what happens when that last developer gets burnt out and quits.
I think AI is being rolled out too quickly, but ultimately I think this wave of AI craze is really exciting to business owners because I think that these people openly think all of their employees are always under-performing because there is a really big disconnect between what business owners want (exponential and infinite growth), and what is realistic. A lot of this is because the business owners, ceo's, managers, aren't actually doing the work, and therefore do not have the tools to adequately and realistically assess productivity, since employees are often just seen as a replaceable burden.
Yeah AI honestly would be best replacing those middle management jobs. AI ironically would be more compassionate and knowledgeable about the workload then actual people who got their job through nepotism.
I get what he was saying, he was talking about the C-suite folks in a start-up or really small company, who are really more like "senior employees who have a fancy title". I worked for a newish company of like 40 people 3 years back and the CTO was constantly writing code and even reviewing other people code and pull requests. Good guy, really smart. It kind of blew my mind to see it at the time. I think all of the C-suites at that company *except* for the CEO were very hands-on with everything they did, but to be fair to the CEO he was retiring like right after I arrived so he kind of knew he wasn't really around for anything.
I think he’s also thinking about C-suite that are also old guard for the company, like the Blizzard boss that Thor was talking about who said anyone could email him basically if the chain of command is protecting bad people. He sounds like the C-suite that you’d want.
C-suite executives are generally very smart when it comes to making the line go up in the short term and then getting a golden parachute for themselves once everything collapses.
I can't imagine CEOs wanting more out of their employees and burning them out. This has never happened before. I can't believe it. Across 0 industries has it ever been tried.
The one rule I've learned about business is that any manager, CEO, or C-Suite person is NOT there based on merit! They're there because they are the grey man who was just competent enough but not too confident to threaten their own boss's job. This effectively means that the world is run by midwits! The average among us! If you stand out too much you'll literally get forced out because you make your "superiors" look bad. The other half of these people are the people willing to f*ck people over just enough to not lose clients or customers but what they're doing is still immoral or illegal in one way or another! There are an extremely small number of businesses or corporations where there's a true meritocracy!
The more experience I get and the more companies I work for, I become even more convinced that the push for "developer productivity" causes massive problems. Yes, some engineers are much better than others, but pushing productivity doesn't help things. Is there such a thing as "mechanical engineer productivity" and "civil engineering productivity"? What about "managerial productivity"? I hope not, because that would be terrible too.
@@username7763Civil Engineer here. Yes, it does exist, and it does suck. The "hide things" or "release it as crappy as it is" measures to reach some artificial indicator is terrible in engineering as well and has lead to quite catastrophic results. And it still happens.
@@ufazig Here engineers have to follow a code of ethics and cannot sign on actions that go against it. The result? The suits get engineers from other countries with no such regulations to sign instead.
Productivity gains still require decision makers to actually understand the problem, the work being done and the solution. Failing to understand all 3 is the baseline in business management.
Unions are dead because those who do work still have huge amounts of competition for their jobs. As soon as the company realized someone or a group of people are in a union they will fire them. Labour is cheap. To much competition, to few jobs.
Yeah and that's illegal! Don't worry the government will not help ever because it was captured during the industrial revolution and the victory for the wealthy is total and complete. Words will not work.
I work on a team of half FTEs and half offshore contractors. My manager has the contractors complete all new features. Why? Because they complete the worst the fastest. They are uninterested in taking the time to writing meaningful unit tests, doing thorough analysis of the changes, or refactoring the code to be SOLID and readable. Oh and all the FTEs are swamped trying to clean up the hundreds of complex bugs they introduce and how long it takes to unravel their spaghetti and fix it.
@@cristianguerra1617 Absolutely, management is fully responsible in my opinion. The contractors have no reason to do any of those things because our management isn't interested in standardizing it and I can't force them to do it without leadership's backing. In fact my manager gets to pump up the numbers for bugs that his team fixed under his leadership this way.
Yup, and it isn't the contractors fault, this is what they are incentivized to do. And this will look good in the powerpoint, allowing managers to show how they cut costs and improved productivity. While the costs are just shifted and increased.
Yes because bosses think it helps us, but we have to train it to help us, and it messes up with customers who then complain to us about it not being trained well enough... it's a lose-lose-lose situation.
The worst thing is that in healthcare, the bare minimum they expect is a herculean effort. I've been saddled with 50 patients before as a nurse for 12 hours. You know how impossible that is. You are supposed to pass bedtime meds in a 2 hour window by going to each room and performing the med admin and doing their treatments. That's 2.4 minutes per person, and some of the treatments take 15 minutes. It needs to be done in the legal window. It just isn't. This doesn't include the documentation you have to do.
I just got laid off from a contractor position for the govt, supposedly they've replaced my function with AI: My job function was redacting safety investigation evidence documents for FOIA public release. Half the content was handwritten and reproduced several times, looking like a scratchy handwriting Captcha. If Captchas are still filtering AI, then how is AI reading my documents? Furthermore, the documents were replete with human error: name misspellings, information split across spreadsheet cells, etc. I just can't believe that AI is able to "see" and catch this kind of unexpected data.
Nothing a 1000 indians in Calcutta can't do "With an infinite amount of Indians and typewriters, you will eventually get them to write Shakespeare" - some guy
Because developers spend more time writing prompts 5000 times and fixing AIs buggy, not quite right code rather than just sitting down and *writing the code in the first place*
Mostly people are too ambitious with what they ask of AI. It can do a zero-shot program up to roughly 40 lines of code, with debugging messages all the way up to 80 lines and if you need to just make small changes, it can usually work with code up to ~200 lines. Beyond these limits it breaks down and you become a prompt engineer, so as long as you use it to speed up / make first drafts it's great. TL;DR As long as you manage the higher level logic and can break your code down into functions, AI is a magical productivity tool par none.
@@MrVohveli Yeah, Ive never been able to make it work for that many lines with anything remotely complex in requirements. Yeah, if I said "make an api for common thing X" it could probably do it. But if I said, "make some thing with X, Y, and Z requirements" it only does it partially right or it loses context and I have to reprompt over again. TL;DR As long as you are willing to repeat yourself over and over and you're making something easily identifiable, then yeah it may work passably
It's more that AI is NOT being used to make their jobs easier.. It's being used to help collect further data on how their jobs are done, so better AI can be trained to fully replace them.
@@jshowao In the XYZ case you'll receive better results if you have it make X first, then ask it to add Y and so on. They are very limited currently, but I don't see why GPT6 and above couldn't generate entire programs up to several thousand lines. I recommend messing around with Autogen for an active debugger, watching it generate, run and debug its own code is quite something.
I've been working on a hobby project to implement a posix style kernel, and I've found it super useful to essentially sift through all the intense detail in data sheets/manual for things like PCIe. It's a massive productivity boost in that regard. I've also had a great time figuring out new language semantics by asking it questions like "how do I make this code more idiomatic? Can I use feature X that I've heard is a good thing in this language" but I've not had much luck unless I'm kind of hand holding it to index the correct information from it's data. It's not so great at "I want you to just kind of make a thing that solves this vaguely defined requirement." It's insanely powerful if you use it right, but it's not a magic wand.
Sure, if you're using AI as "better Google" then it can be very helpful. It's pretty good at interpreting broad ideas and rewriting them in whatever style you desire. The problem is that the kind of companies this article is targeting think you can ask ChatGPT questions like "write a posix style kernel for me" and then insist that your "productivity" should be 40% higher based on that overly optimistic misunderstanding of the technology.
That's exactly how I've been using it. As a research assistant, but not a code replacement tool. I rarely ask it to write code for me, but if I do, I always make sure I understand what's it doing before using it, and almost always hand-type it into my project as a means of doing so.
Over the decades I do see teams incapable of absorbing new tech. This is a "medicore" to "disfunctional" org problem. If it's not your org, bring the doughnuts to work and be thankful. Most places have things holding them back. Superstars are expensive.
This is combination of “a lot of people just don’t know how to use AI quite yet”and “workplaces that are already productive aren’t rushing to throw AI in the mix”
@@SeeAndDreamify AI is a broad field with many tools/techniques. I think you're right to say certain AI applications aren't ready to be used in particular ways, and vice versa, but being too general risks people interpreting AI as some big monolithic thing that will eventually reach their sci fi expectations one day. Arguably it's also the same problem: using AI correctly means knowing its limitations- using AI can fail not because it isn't ready, but because a business thought it was ready or even relevant to their process.
@@tukib_ "but being too general risks people interpreting AI as some big monolithic thing that will eventually reach their sci fi expectations one day" Which people already do. AGI is just around the corner, 3 months top baby!
I’m freelance. We don’t disappear on long lunches, gossip over coffees. We are never late, we work extra hours to finish projects on time, we don’t call in sick. If there is a zoom meeting, I just listen and keep working on what I can. I try to cheer up the people who feel stuck there on salary.
Once a boss somewhere brings in a freelancers/consultant they are deemed to be making an investment by their superiors. They do so because the task is deemed important for their career. Now it this external is a critical component in furthering their career. Bosses won't let impediment s blocking their career progression remain for long. Actal employees on the other hand are not treated like this because they are business as usual. As long as there is nobody yelling too loudly about those fires in the hallway they can be safely ignored.
Also, we've been working longer hours post industrial revolution. Our farmer ancestors worked a lot less. Heck they even had siestas! So are you surprised?
It's to competitive. You can't have a union if there's someone who's willing to compete who isn't part of it. Companies will always hire people who aren't in the union over those who are.
GPT wrote me a script to invert PDF books so reading them won’t hurt my eyes. It doesn’t work, obviously but it’s cute that it tried. And I have some cool new PDF CLI tools. No idea how to use them
I forget what terminal PDF reader I have (I think it's zathura?), but it has vim motions and color inversion as a motion. Anyways, I literally never use it, because I practically never read PDFs.
Farming for your family doesn't require that much time and effort. However, if you're farming for a civilization that consists of a significant population of non-farming citizens then it does require much time and effort and thus requires automation or economies of scale. Non-farming citizens are doctors, veterinarians, lawyers, teachers, builders, politicians, writers, artists, merchants, etc.
Plus subsistence farming is only super busy during sowing and harvest season, where so much preparation and processing needs to be done, of course also the occasional pest / natural disaster in the mix in worst case scenarios. The rest of the time it's just mostly watering and weeding. In some locations subjects even have wine bread and cheese provided by the fiefdom, so full self sufficiency wasn't even required lol
That massively depends on what you're farming. Managing a herd of dairy cattle is a full time, seven days a week job. Even subsistence farming for just your family will involve vegetables, dairy, meat, etc. Preparing, sowing, maintaining, and harvesting crops. Feeding animals, collecting eggs, milking, and slaughtering and butchering animals. Anyone who thinks that's easy has probably never stepped foot on a farm.
What, how do you go from 99% of your population being farmers to almost nobody being farmers without those 99% being necessary for the job? Were the aristocrats really just letting all their serfs do almost nothing for all those millennia?
The scale of modern farming isn't what he's talking about. In the past people had a lot more free time than you realize. They worked hard when it was required, but it wasn't required half the time.
On the topic of salary amount before or after taxes, in my country all advertised salaries are “after taxes”. As an employee, my employer pays my taxes, public healthcare and retirement/pension rate. I get a digital payslip each month so I know exactly how much they paid and I don’t have to worry about paying taxes. The state sends me a digital report each year telling me how much Ii’m expected to pay vs how much has already been paid, in case there is a discrepancy. So far I’ve only had to click the “I’ve checked and agree this is correct” button… once per year. Why the hell would I want to do the tax math/payment myself in 2024, when it’s all automated for me? I guess it’s an American thing or maybe something in the more developed west.
In the USA, the taxes you pay on income are not known at the time of earning it. They depend on your total income. It isn't a fixed percentage, but a progressive rate. And then there are deductions which reduce your taxable income, some of which you don't know what you'll have until the end of the year. You estimate how much you think you'll owe and your employer sends that estimated amount to the government. Depending on if your estimate is high or low you'll get a refund or have to owe taxes and penalties sometimes too. Some people also have to mail the government a check periodically to avoid being too far negative at the end of the year. Bottom line, your employer has no idea what your after-tax income will be.
that is true of a noob junior developer as well. If you use Ai on a project to learn it actually teaches you as you build as long as you try and understand what is going on and read up about other ways of doing it. But someone once said of it works, it works.
I think also freelancers are hired when you need a project with their exact expertise, so they know a ton about that niche. If your normal employees don’t know as much about that area of development which is why you might hire a freelancer, they might not perform as well as the freelancer doing that one thing.
You have to a plan for who will take it over from the freelancer. The same problem that leads you to outsource it still exists when the project is done. The freelancer will go onto the next thing and you then have a project that you cannot support, extend or maintain. I saw this with one company I worked for where FPGA design was outsourced while internally no one knew FPGAs. We ended up with a ton of hacky workarounds because no one would touch it. It was actually a pretty simple FPGA design that anyone with a basic knowledge could have handled. But no one did.
Massively underrated comment. Companies are trying to sell "AI-powered" products only because some sickeningly overenthusiastic, half-informed marketing exec read about it in one of the countless business advisor magazines (because those are always in touch with what the people want, right?), without any regard for the maturity and usability of any of it.
Nobody will be talking about AI 5 years from now… this AI talk has been happening for the past 60 years. Btw, we don’t even have a definition for AI… it’s simply a corporate talking point.
This. Sad to say that even in this day and age, illiterate and poorly educated people exist despite the internet. Now, imagine showing these people LLMs. They'd think it's Skynet made real.
Well they're outspooking the average. Its just we're figuring out the average or below average of humans all are in management. Its their jobs that would be most effective to replace. However in clown world its all the specialists being laid off.
We have been leveraging machine learning since before the AI explosion. The large language models are not all of AI, and expert systems are a completely different animal. It's true that employers will always squeeze more out of employees if they can, but we can have the computer do the tedious bits and let humans do what humans do better.
I would like to point out the most (socially) competent people in charge are the ones who know they are bullshitting you non-stop, and also know everyone will smile and nod. They also know ultimately if you just bullshit everyone and then somewhat let people do their things while paying them, solutions start appearing because the whole exercise is just social etiquette, yes, they know you speak ill about them at their backs, don't even think for a moment each part doesn't know exactly what the other part is thinking about the exchange. Just to clarify, this isn't vranyo (and thanks to Putin's SMO for giving us this word), the difference between vranyo and bullshitting is with vranyo neither part expects anything to be accomplished (you pretend you work for us and we pretend we pay you), meanwhile western executives, while bullshitting you, still expect things to work, that's why western societies somehow are still stomping forward despite the manias and dysfunctionalities while the Global South needs to pay wumaos, bots and metodichkas to fool disaffected westerners into thinking there is a place they can flee from the bullshitting. There is no place.
The problem with most people using AI is that unless you understand the requirements and can communicate them unambiguously, and then understand enough of the answer to verify it, then you are going to to hit the 'garbage in, garbage out' problem.
I love how this thing talks about innovation increasing 2x. They empirically measured everything unprecedented. And guess what. It all turned out to be quantifiable with a single, convenient measurement. And it turned out we've got twice as much never-before-seenness in our whatsits. Great news.
Your point about job performance is SO spot on it's almost criminal. At my previous company I worked my ass of, I actually founded an entire new business area. But then my dad got cancer and some other things happened and I "underperformed" that week when he was in the hospital (he also got blood poisoning) and voila, all of a sudden I got those "why are you not performing" bs comments. Unbelievable. Fortunately I found a new job and switched and it's currently great, but still. Really sad when you give so much to a company and they don't give a shit.
Yes they don't care, and upward movement is a lie, brown nosing is the only skill needed to go upwards, easier to move to a better job in another company. Also your just wasting your life working, if one has no debts then why do we need to work.
As an artist, I have to thank artificial intelligence. It has increased my work and there are many people who want to learn drawing after using artificial intelligence XD
that's why i don't fear technology. if it makes things better, now you have more work to implement the new thing or if it solves a problem, then now some things that were not possible before are possible now... and the end is more work some jobs are lost... but the amount of work always goes up, it will never stop going up
Pipedream: technology will decrease workloads and give people more free time! Reality: Here's 500 more things to work on... and yes, if this looks like the existing work other teams would've worked on, that's because it actually is. NO 15 HOUR WORK WEEK FOR YOU! Good luck! **Looks around and realizes that no other team exists in the office.** _New Objective: SURVIVE_
That's because right now, AI is NOT being used to make their jobs easier.. It's being used to help collect further data on how their jobs are done, so better AI can be trained to fully replace them.
Freelancers are better perceived performance because full time employees are asked for everything they can give the employer, and freelancers have billed hours. They pay, they get. Full-time workers are used to the business and know the company but that is not valued at all. They are thrown under the bus all the time.
What I'm hearing is from this article is "your employees are burned out by your AI-driven expectations, therefore you must hire some highly trained free lancers that will set up a castle of AI-native processes, and then close the contract as soon as your expectations are no more economically advantageous to them, leaving you with even more unhappy employees".
They should ask the solo developers and small teams how AI affects them. People who don't have managers that increased expectations, who personally chose to include AI in their workflow, and designed their pipeline around it.
I work 40hrs for my salary. If I somehow need to work more, I "charge" the standard overtime rate via the time reporting system. Wages and salaries do not work the same all over the world... :)
I know a guy who had 3 contract jobs at once and anytime he was drowning in work he just sub contracted someone from India, he did say that didn't happen that often.
Its called hallucinating. Its when data is pulled from dissimilar subjects relating to similar keywords. Ai cannot verfiy its own information and spits out bs chatbot garbage. Most actual people do this too, its not just ai. This world needs to recognize the limits of stupid low iq peopleand that they will never be productive, just robotic.
Remember a while back my comment "AI = Productivity goes up, Quality goes down". Well I was wrong. It looks like I have to update: AI = Productivity goes down, Quality stays down
Summary: A study shows that 77% of employees report that AI has increased their workload and hampered productivity, while 96% of executives expect AI to boost productivity. There is a disconnect between the high expectations of executives and the reality faced by employees, who feel overwhelmed and are experiencing burnout. The discussion around the use of freelancers is seen as potentially biased, as they may be more productive but often provide short-term solutions. To fully leverage the benefits of AI, companies need to fundamentally rethink their work models and integrate alternative talent pools. Overall, it is emphasized that the integration of AI into the workplace presents both opportunities and challenges, making it crucial to develop a realistic understanding of its impacts.
If you think about it, using AI is like being a coach for a team of both geniuses and complete knuckleheads. Your main job? Yelling at them to get the simplest tasks done!
4:56 this is so true at companies. I’ve made this mistake before. You start off never taking PTO and skipping your lunch. It becomes harder to reel it back afterwards. Precedents are important to set
To clarify, AI didnt fade out and come back in the last 50 years due to algorithms, but due to compute. in the 1970s we didnt have enough compute. Right now we are slowly getting close to the estimated compute of the brain, back then this was exponentially far away. To think there will be another AI winter because of the last AI winter just shows poor understanding of the subject.
@@mage3690 "logarithmic curve" is actually very non-descriptive. If you understand that all you need for intelligence is scaling up the model components, so either architecture, number of parameters or number of quality tokens, you basically get more than most people including prime himself. Scale is all you need. If you increase compute you allow for more parameters, longer training aka more quality tokens and through research, the only part prime knows, you get minor improvements to the efficiency. But nothing you cant get with more compute.
Hey Primetime, hope you see this. One thing you have to consider about freelance or temp workers compared to full time staff is the implied job security. Most perm workers live under the assumption that they have their job and all the benefits/safety nets it comes with, like consistent health insurance included, labor protections (I.e. unemployment, COBRA, etc.). freelancers don’t have that, so they’re able to demand a higher wage to make it worth the time, loss of those benefits, downtime between freelance contracts, and relative job uncertainty
I just wanted chatgpt to do some boring 20 lines copy paste pandas code, first suggestion, revert the entier dataset, do calculations and revert it back. Suggestion two, only revert it once and have two copies, suggestion three, just use shift (with a wrong amount), and halusinating it would work with an output example. Ye, it took longer than just writing the code myself (heck, opening chatgpt and asking probably took longer), and it didn't even end up giving one good solution. And I rarely ever try anymore, the solutions are trash, and most things doesn't work. Its quite alright for very short basic problems, but its also not something most programmers need much help with.
Point #4 basically says instead of planning projects as a set of specific jobs that you need done to get the project done (I need someone to get me coffee every MWF at 10:00 AM) to skill-based (I need someone who can walk fast and handle hot substances). To add AI to the mix that would be an employee that can program a VoIP AI to call you around 10:00 AM every Monday Wednesday and Friday asking you how you'd like your coffee, take your input, and map your input to an API call that orders a coffee with your preferences and have someone deliver it to you. Insane and stupid, but scalable.
Problem is not all AI is the same. Claude 3.5 is producing very good code. Google Gemini? Not so much. Python, Java, or C++, HTML, JS, they are very good at. Others, seem to be very hit and miss. I expect this to only get better rapidly judging on how much better the last year has given us.
We started paying for Copilot & Chat GPT around January of 2023. A large chunk of our current work is completely refactoring work created with the assistance of machine learning tools. Most of the markup is fine, map/reduce/filter/sort style functions are fine, but anything that required implementing a solution vs autocompleting a formula has ended up with tightly coupled code, that is very slow for large operations. The ML stuff is fine for small & narrowly defined functionality, but it does not work at all for enterprise level software in the long-term (at least not so far) We still pay for ML services, but frankly, I'm spending more time reviewing and rejecting code as a lead than ever. On one hand, I review commits a lot earlier in the process than I used to, to avoid bad code (I'm not sure, is that actualy a good thing or a waste of time?). On the other hand there's so many more "high level" issues, that people are continously re-doing stuff. Frankly, the box has been opened. Devs (on average) will allways take the easy way out, which now means using ML code without thinking it ober. Quality control is going to become the big differentiator in the future. I don't know If that's sustainable, I'm personally mentally exhausted.
I don't work in tech, but I work as an artist in gamedev this holds up. One project started implementing ai in the pipeline, the amount of artists working on this project was significantly decreased yet deadlines became tougher (like a task that beforehand was estimated to last 40h now was estimated at 16h) Artists who had to work on that project complained a lot, because although ai did make the process faster, but not by that much, there were still plenty of mistakes that had to be fixed after the ai and some tasks it did downright terribly. Artist were very overworked for the same pay as before.
To the freelancer segment, I would also add 2 things : - highly paid free lancers are hired for their expertise, full time employees not always, you might be hired to code but have to do powerpoints instead - freelancers participate to the estimations because their business depends on it. In some companies, Johnny the manager who has not coded in 6 years does the estimations for bob the coder, and then bob does not meet the deadline
AI is even affecting basic retail worker's jobs. They set it to organize the tags on the shelves, and it doesn't even understand counting. There is a system and there are sections and levels to the numbering system, but now they'll be in the wrong order, in the wrong sequence, going backwards, and organized so that products dont fit in the first place
India is the most common place for outsourced remote work for medical billing (and we've seen them used for "AI"). Malaysia is also common. They oft have Morning/Night shifts too, so timezones aren't a major issue for them. Instead, their ridiculous amount of holidays and minimal competent staff is more likely the issue. (edit: their more competent staff doesn't usually do the remote/cheap labor stuff. it's just a numbers thing) (also, the whole "before or after taxes" was pertinent for your question. places like the UK deduct and file the taxes for you, so you really only care what you make after tax)
the thing about subsistence farming is you can only do so much before you just have to wait for things to grow, so no, people who were subsistence farming were not working all day every day. and they stopped working when the sun went down, electric lighting increased the average workday significantly
The great thing about ai is that people are finally shutting up about blockchain.
Until ai block chain comes out
@@LongJourneys Ai native block chain
True, were it not for Bitcoin bros.
My native-AI blockchain cloud dynamically assigns networking priority via lootbox mechanics for efficient handling of bandwidth normalization.
@@BalkanSmurf How can I invest?
I’ve noticed a correlation between increased workloads and employee layoffs. Essentially, companies seem to be replacing 50% of their workforce with AI and expecting the remaining employees to take on the additional responsibilities of those who were let go.
Tbf that's exactly what productivity means - with technology one farmer can now do work that once took hundreds of labourers. Problem is that right now the landowners have decided their new magic ploughs and sickles are twice as good so half the peasants can be used as target practice and the rest are expected to bring in the harvest with half as many hands. Workload obviously goes up when the masters demand more work from fewer people on less pay. And they'll seize any reason to demand it.
@@thesenamesaretakenAny master is alive and can be made not
This is a problem with not letting companies fail when they make poor decisions. We need these companies to fail so that these middle managers and executives don't keep piling up and adding parasitic undergrowth. And then you let new companies take their spot that actually make good decisions.
@@dmitriyrasskazov8858 Names, addresses, etc.
people kinda suspected that was what was going to happen
Fire 50% of their work force, then wonder why the remaining employees are complaining they're being worked to the bone.
I never worried so much about losing my job to AI, I always worried management would listen to some 20 minute lecture and be fooled into downsizing, thinking it was the solution to all their staffing problems. Imagine my surprise, seeing it unfold in real time.
EXACTLY
Wonder? They know and don't care.
@@BaronCreel The thing is, they can't "not care" forever, it's just not how things work. Sooner or later they have to do something.
@@Cye_Pieit does work though, these managers have golden parachutes to get into another company and they know it, all the fallback and consequences go to the junior staff/ long term employess outside of management, they can afford to not care with connections, it's not that they are incompetant, they know they can get away with it and repeat it somewhere else
And as long as the remaining simps keep working, they will continue to do it.
My mom is a tax professional working at a multinational corporation. The job itself is stressful enough. She was in a department of 5 where they had this fresh grad just doing data entry. The kid was great, I met him a few times. Just sits there quietly, doing his work. 99.5% accurately from what I've been told. Some AI startup came along, promised to eliminate his position and many others, management jumped on it and the AI is absolute dog doo doo. It's only about 10% accurate and costs more than the kid alone lol. The corporation has decided to stop using the AI but refused to hire on a person which makes zero sense. My mom who hasn't done data entry in decades is now stuck doing Jr level work, while also being expected to complete her own duties. This **** is ridiculous. She is about a molecule away form walking out and I can't wayt for that day. They'll have to hire at least 2 people at her salary each to get the same amount of work done. I saw job postings from them but didn't bother applying. F em.
Yeah AI doesn't belong in any area requiring 100% accuracy. This is a well-known limitation. It is great when something that looks right and is mostly correct is what is needed. You'd expect an AI startup or your company management to know this. It isn't a secret or anything requiring deep knowledge.
Your mom should be looking for a new job
And when she finds it she should leave without notice
@@matheussanthiago9685 Moral here is that she got fired but the plan backfired. She is still getting fired. She just hasnt realized it yet.
@@matheussanthiago9685 And convince her to leave in the middle of a very important project :)
Yeah she just shouldn't do it.
As someone who has worked at several tech companies, I can tell you those years were unbelievably unproductive. I was worked into oblivion, don’t get me wrong, but I was working on a whole lot of nothing. I was expected to learn countless pieces of software that were supposed to help us stay organized, make us more efficient, and that kind of thing. They never did. Just made work more difficult. No one ever followed the process. Everyone pretty much just made their own rules. I remember I was using a project management software at one company and at one point another department decided to use a different project management software. We then had to learn their project management software because we always worked with them. Instead of building out projects in one piece of software, now we had to build it in two. I had to check off tasks in two softwares and oh by the way, had to send an email and a slack message on top of that to our stakeholders anytime a project was finished, then I’d have to monitor all 4 channels to capture any feedback. It was just absolutely ridiculous.
Ohhhhh, feeling like that now. Between Microsoft teams, email, airtable and a "custom software" for our company, I, the human, am ensuring the same information existing in like 5 formats for different people and me making sure they all stay the same along the way with this software that was suppose to streamline the process, but doesn't.
Bad management is the problem. Listen to your employees people, you hired them for their expertise!
But machines do not talk back!
File this under "stuff that will never happen".
@@T1Oracle But why, when we can chase buzzwords for guaranteed success and have fingers in our ears to all those boring complainers? Like don't they realize they are killing our mood with all their reasonable concerns?
Unless you hired interns and cheapest interview-pleasing people who have no expertise, knowledge or idea what they should be doing
Nah. Slaves should not talk
My org is trying to implement AI to replace humans, upper-management has blocked off negative feedback about AI so that they can continue overpromising so that C-Suite doesn't hear anything except that it's cutting costs. We're all hauling ass working overtime to try and keep up with the promised speed of AI. It's like the car in flintstones where it's really just a few dudes pedaling for their life.
Don't haul ass
Make it fail
Why the fuck are making an effort to fuck yourself over?
You all need to collectively stop
Well you know, you could all just stop, management will get mad, but after a certain point executives will notice and come down to see the commotion
AI = Actually Indians
You are pedaling for someone elses wealth.
ironically, the best job AI is suited for is management. imagine if we didn't need work our assess off to pay the salary of a CEO
CEOs don’t make that much in comparison to 1000s of employees
@@Artifex_Prudenson average a CEO makes as much as 344 employees, it's definitely the best use of AI
@@dudu28r81 you don’t seem to understand the value of certain individuals with true vision. Not all individuals are equally competent
@@Artifex_Prudens sure you can argue that but they used to be paid MUCH less(orders of magnitude) and were doing just fine. In comparison a lot of the working force is one step away from being homeless
@@dudu28r81 then cease the means of production and do better yourself
There is one major reason why freelancers are more efficient than full-time employees, freelancers don't have useless, and knowledgeable management and higher-ups implementing wasteful, redundant, and pointless policies onto their employees. I worked a retail job where my certain department was so small, just me and another guy, that they removed the Manager position for our department. Everything immediately improved, because the people responsible for making the big decisions where the people doing the actual job
I moved from an amazing workplace where our manager was maybe 40% manager and 60% on the floor working, to a workplace where the manager (same job) was only manager and had no idea what we were doing. It was so bad. He didn't know about safety, how long things took, what we were even doing or talking about. Everyone were complaining and eventually I quit.
Tech has a massive, massive, systemic pattern of just throwing the newest thing at every problem when it wasn't designed to solve every problem and in fact can make many problems harder to solve, not easier. It's very odd to me that with so many so-called "smart" and "genius" people that we have, we all collectively do this over and over and over and expect different results.
the geniuses aren't in management positions
See it's less that tech people are actually smarter and more that "being smart" is intrinsically tied to their sense of self worth, so they will often pile onto the thing that makes them seen as smart 😪
many are of course smart, but not too much smarter than law or finance or medicine people
i mean, if they receive millions in venture capital funding to use a hammer to drill screws, then i would say the venture capitalists are the idiots, not the startups
@@spwashi underrated point
@arkie87 why not both?
It's funny that managers and companies will complain about employees not coming back to face to face work but then hire a freelancer whom they never even see and could be in an entirely different country!!! WTF!!!
Because it's usually about 10x cheaper. So they can hire 3-5 people instead of one locally and save half their salary, even though those 3-5 people will do shitty work.
When Autocad first started it was considered revolutionary in respect to drawings and drafting. It was expected to reduce the workload for draftsman as replication and copying was greatly simplified. As exactly mentioned, expectations on what can be produced sky-rocketed at the same time.
Simplify work -> Get a higher workload because of that
The cotton gin was created to lighten the workload of cotton slaves, but only cemented their value to the slave owners
It reduced the workload so much draftsmen barely exist anymore
Tale as old as time
@@jacobmansfield-go9fz Or they're now called something else? Designers.
About freelancers and productivity:
as a freelancer you can afford to not give crap about quality. You will not have to maintain that code for a long time. And when that code explodes or gets expoited, you are long gone.
Yes, this means you can be negatively productive where your code requires more time to fix than it took to write it. As a field, we don't need more code. We need code to be correct. We need code to be simplified. But we really don't need more KLOCs.
Pretty much. Like a cum and dump relationship, but in programming.
@@username7763its not a freelancer faults tho. We get hired for specific task not knowing or having time to dig deep into the software. And free market is cheaper and faster the better.
And then we tend to not come back to same project because next person is now cheaper than we.
Full time is always better than part time. Part time is good for temporary support but needs to lead and managed by ppl working on a project.
We just lost our 4th freelance dev we relied upon. That can hit you back too. But then again we never had enough work to sustain him full time or even part time.
Company we deal with lost their full time and instead of hiring new full time they rely on freelance to do the same that this person been doing full time for years.
@username7763 yes but if a company is going to pay me the minimal ammount im only going to give them the minimal quality
that is called a scam artist. Where I'm from once you develop it you support it on a SLA retainer. Any free lancer that develops and disappears is a >on, you develop with the aim to get more SLA and recurring income.
On the point of farmers working 6-7 day workweeks, they also typically stopped working when the sun went down and during winter (in colder climates) not to mention not everyone was a farmer at the time
They also took a nap if necessary.
Work for 4 hours, finish up, do some house chores, is not the same as a warehouse telling people to stand around and push boxes for 12 and thrn go home, make their own food, chores, etc
Also, a lot of time on the farm is waiting. Waiting for the time to do something.
The crops need time to grow. Farmer can only do so much.
There are other tasks but not so much
@@huntermad5668friend sent me recently that at certain point peasants worked 150 days a year as church wanted them to have some rest. Go figure.
I love when management makes decisions based on an article they read on a plane. Here is that article.
The CEO of one of the biggest Nordic traveI agencies canceIIed a very much needed e-commerce website and database upgrade project and diverted aII resources to instead focus on "APIs" because he had heard about something caIIed "APIs" on a conference. That project faiIed. Many many miIIions down the drain. Situation got so fubar that the shareholders had to seII the company roughly a year later, and thus have the CEO fired. Ironically, he was repIaced by the previous CEO of the company, that's how bad he was.
@@GackFinder wtf how do you even invest in APIs?
@@Daniel_WR_Hart That's one of the many questions that we (architects and developers) relayed to middle management, whom in turn completely ignored those questions in discussions with the C-level managers.
I should start scamming CEOs.
@@KidCorporate I was thinking of creating an AI project called Quentin that can replace a CEO and pitching it to VCs
Oh wow! who imagined that the only developer left would take the work of the 5 fired for AI-native?
Honestly I still don't get what "AI-native" means. It probably isn't hard to figure out, my brain just has a "yuck" reaction to that word combination and refuses to get it.
@@the-answer-is-42 no, it's because that actually doesn't mean nothing. It's just a buzzword to get VC's money.
@@alexandrecolautoneto7374 Yeah, probably. I just don't like to dismiss things I don't understand as nonsense before I look into them, and I'm not exactly motivated to look into "AI-native".
@@the-answer-is-42 It means "copy and paste from ChatGPT's reinterpretation of a stack overflow answer instead of copying directly from SO".
> who imagined that the only developer left would take the work of the 5 fired for AI-native?
The real question is what happens when that last developer gets burnt out and quits.
I think AI is being rolled out too quickly, but ultimately I think this wave of AI craze is really exciting to business owners because I think that these people openly think all of their employees are always under-performing because there is a really big disconnect between what business owners want (exponential and infinite growth), and what is realistic. A lot of this is because the business owners, ceo's, managers, aren't actually doing the work, and therefore do not have the tools to adequately and realistically assess productivity, since employees are often just seen as a replaceable burden.
Yeah AI honestly would be best replacing those middle management jobs. AI ironically would be more compassionate and knowledgeable about the workload then actual people who got their job through nepotism.
I wish I could have your optimism that C-suite folks, especially in tech, are smart people who deserve to be there.
I've yet to find any evidence of this.
I get what he was saying, he was talking about the C-suite folks in a start-up or really small company, who are really more like "senior employees who have a fancy title". I worked for a newish company of like 40 people 3 years back and the CTO was constantly writing code and even reviewing other people code and pull requests. Good guy, really smart. It kind of blew my mind to see it at the time. I think all of the C-suites at that company *except* for the CEO were very hands-on with everything they did, but to be fair to the CEO he was retiring like right after I arrived so he kind of knew he wasn't really around for anything.
I think he’s also thinking about C-suite that are also old guard for the company, like the Blizzard boss that Thor was talking about who said anyone could email him basically if the chain of command is protecting bad people. He sounds like the C-suite that you’d want.
C-suite executives are generally very smart when it comes to making the line go up in the short term and then getting a golden parachute for themselves once everything collapses.
@@Tobi-ci3nsno, no they aren't. They think they are though.
I can't imagine CEOs wanting more out of their employees and burning them out. This has never happened before. I can't believe it. Across 0 industries has it ever been tried.
😂😂😂😂
Increase in productivity never translates to people working less, just less people working
This article really nailed that Chat gpt writing style.
I mean if you want to sell AI-native freelancers, you ask one to write a Forbes article and he uses AI.
The one rule I've learned about business is that any manager, CEO, or C-Suite person is NOT there based on merit! They're there because they are the grey man who was just competent enough but not too confident to threaten their own boss's job. This effectively means that the world is run by midwits! The average among us! If you stand out too much you'll literally get forced out because you make your "superiors" look bad. The other half of these people are the people willing to f*ck people over just enough to not lose clients or customers but what they're doing is still immoral or illegal in one way or another! There are an extremely small number of businesses or corporations where there's a true meritocracy!
> Rethink how you measure productivity...
So, they're saying they're more productive, and they've redefined productivity? Ya don't say.
smoke and mirrors everywhere in this article. yup
The more experience I get and the more companies I work for, I become even more convinced that the push for "developer productivity" causes massive problems. Yes, some engineers are much better than others, but pushing productivity doesn't help things. Is there such a thing as "mechanical engineer productivity" and "civil engineering productivity"? What about "managerial productivity"? I hope not, because that would be terrible too.
@@username7763Civil Engineer here. Yes, it does exist, and it does suck. The "hide things" or "release it as crappy as it is" measures to reach some artificial indicator is terrible in engineering as well and has lead to quite catastrophic results. And it still happens.
@@ufazig Here engineers have to follow a code of ethics and cannot sign on actions that go against it. The result? The suits get engineers from other countries with no such regulations to sign instead.
I see the shirt says 'Twitch,' but all I think is 'FedEx.' They have a mental monopoly on black shirts with purple accents.
but taco bell
I think that is just an American thing 🤷🏽♀️
Productivity gains still require decision makers to actually understand the problem, the work being done and the solution.
Failing to understand all 3 is the baseline in business management.
People are working less in the sense that workforce participation has been steadily dropping since the 90s. Working people are working more.
I bet things get so painful that unions become a thing again, despite all the dei
Unions are dead because those who do work still have huge amounts of competition for their jobs. As soon as the company realized someone or a group of people are in a union they will fire them. Labour is cheap. To much competition, to few jobs.
Yeah and that's illegal! Don't worry the government will not help ever because it was captured during the industrial revolution and the victory for the wealthy is total and complete. Words will not work.
I work on a team of half FTEs and half offshore contractors. My manager has the contractors complete all new features.
Why? Because they complete the worst the fastest. They are uninterested in taking the time to writing meaningful unit tests, doing thorough analysis of the changes, or refactoring the code to be SOLID and readable.
Oh and all the FTEs are swamped trying to clean up the hundreds of complex bugs they introduce and how long it takes to unravel their spaghetti and fix it.
As bad as the offshore can be, it also sounds like there's no framework in place to guide development
As bad as the offshore can be, it also sounds like there's no framework in place to guide development
@@cristianguerra1617 Absolutely, management is fully responsible in my opinion. The contractors have no reason to do any of those things because our management isn't interested in standardizing it and I can't force them to do it without leadership's backing. In fact my manager gets to pump up the numbers for bugs that his team fixed under his leadership this way.
Yup, and it isn't the contractors fault, this is what they are incentivized to do. And this will look good in the powerpoint, allowing managers to show how they cut costs and improved productivity. While the costs are just shifted and increased.
sounds like a shit tech lead you have there.
Yes because bosses think it helps us, but we have to train it to help us, and it messes up with customers who then complain to us about it not being trained well enough... it's a lose-lose-lose situation.
But but but guys at the Golf club said it’s super effective!
@@errrzarrr it certainly makes me want to club a few things multiple times per day...
The worst thing is that in healthcare, the bare minimum they expect is a herculean effort. I've been saddled with 50 patients before as a nurse for 12 hours. You know how impossible that is. You are supposed to pass bedtime meds in a 2 hour window by going to each room and performing the med admin and doing their treatments. That's 2.4 minutes per person, and some of the treatments take 15 minutes. It needs to be done in the legal window. It just isn't. This doesn't include the documentation you have to do.
You need that tube the court stenographers have so you can just speak your documentation while working 😂
I just got laid off from a contractor position for the govt, supposedly they've replaced my function with AI:
My job function was redacting safety investigation evidence documents for FOIA public release.
Half the content was handwritten and reproduced several times, looking like a scratchy handwriting Captcha. If Captchas are still filtering AI, then how is AI reading my documents?
Furthermore, the documents were replete with human error: name misspellings, information split across spreadsheet cells, etc. I just can't believe that AI is able to "see" and catch this kind of unexpected data.
I’ve done some computer vision work. Sounds like some vendor lied to the government
It can't
At the end of the day
It's just an Indian replacing you
It doesn't but it hallucinate well 😂
Nothing a 1000 indians in Calcutta can't do
"With an infinite amount of Indians and typewriters, you will eventually get them to write Shakespeare" - some guy
Because developers spend more time writing prompts 5000 times and fixing AIs buggy, not quite right code rather than just sitting down and *writing the code in the first place*
Mostly people are too ambitious with what they ask of AI. It can do a zero-shot program up to roughly 40 lines of code, with debugging messages all the way up to 80 lines and if you need to just make small changes, it can usually work with code up to ~200 lines. Beyond these limits it breaks down and you become a prompt engineer, so as long as you use it to speed up / make first drafts it's great.
TL;DR
As long as you manage the higher level logic and can break your code down into functions, AI is a magical productivity tool par none.
@@MrVohveli Yeah, Ive never been able to make it work for that many lines with anything remotely complex in requirements. Yeah, if I said "make an api for common thing X" it could probably do it. But if I said, "make some thing with X, Y, and Z requirements" it only does it partially right or it loses context and I have to reprompt over again.
TL;DR
As long as you are willing to repeat yourself over and over and you're making something easily identifiable, then yeah it may work passably
It's more that AI is NOT being used to make their jobs easier.. It's being used to help collect further data on how their jobs are done, so better AI can be trained to fully replace them.
@@jshowao In the XYZ case you'll receive better results if you have it make X first, then ask it to add Y and so on. They are very limited currently, but I don't see why GPT6 and above couldn't generate entire programs up to several thousand lines.
I recommend messing around with Autogen for an active debugger, watching it generate, run and debug its own code is quite something.
the trick I do is I copy and paste the same code back into the prompt and tell it the error I got. It works to fix the error.
I've been working on a hobby project to implement a posix style kernel, and I've found it super useful to essentially sift through all the intense detail in data sheets/manual for things like PCIe. It's a massive productivity boost in that regard. I've also had a great time figuring out new language semantics by asking it questions like "how do I make this code more idiomatic? Can I use feature X that I've heard is a good thing in this language" but I've not had much luck unless I'm kind of hand holding it to index the correct information from it's data. It's not so great at "I want you to just kind of make a thing that solves this vaguely defined requirement." It's insanely powerful if you use it right, but it's not a magic wand.
yeah, it increased my productivity, but maybe 15%, maybe even less than that.
Sure, if you're using AI as "better Google" then it can be very helpful. It's pretty good at interpreting broad ideas and rewriting them in whatever style you desire.
The problem is that the kind of companies this article is targeting think you can ask ChatGPT questions like "write a posix style kernel for me" and then insist that your "productivity" should be 40% higher based on that overly optimistic misunderstanding of the technology.
That's exactly how I've been using it. As a research assistant, but not a code replacement tool. I rarely ask it to write code for me, but if I do, I always make sure I understand what's it doing before using it, and almost always hand-type it into my project as a means of doing so.
Over the decades I do see teams incapable of absorbing new tech. This is a "medicore" to "disfunctional" org problem. If it's not your org, bring the doughnuts to work and be thankful. Most places have things holding them back. Superstars are expensive.
Are you sure? I think we should really have an ai evaluate some performance metrics so that we can be totally sure its actually happen.
This is combination of “a lot of people just don’t know how to use AI quite yet”and “workplaces that are already productive aren’t rushing to throw AI in the mix”
could also be a bit of "AI just isn't ready yet"
@@SeeAndDreamify AI is a broad field with many tools/techniques. I think you're right to say certain AI applications aren't ready to be used in particular ways, and vice versa, but being too general risks people interpreting AI as some big monolithic thing that will eventually reach their sci fi expectations one day. Arguably it's also the same problem: using AI correctly means knowing its limitations- using AI can fail not because it isn't ready, but because a business thought it was ready or even relevant to their process.
Even when you know how to use ai it's practically useless
@@mattymattffs It is useful in some ways. Mainly in the same ways that search engines used to be useful before search engine optimization ruined them.
@@tukib_ "but being too general risks people interpreting AI as some big monolithic thing that will eventually reach their sci fi expectations one day"
Which people already do. AGI is just around the corner, 3 months top baby!
Freelancers aren’t bogged down with maintenance tasks, ad hoc requests, meetings.
Freelancers focus on a project. Thats why they move things ‘faster’
I’m freelance. We don’t disappear on long lunches, gossip over coffees. We are never late, we work extra hours to finish projects on time, we don’t call in sick. If there is a zoom meeting, I just listen and keep working on what I can. I try to cheer up the people who feel stuck there on salary.
Once a boss somewhere brings in a freelancers/consultant they are deemed to be making an investment by their superiors. They do so because the task is deemed important for their career.
Now it this external is a critical component in furthering their career. Bosses won't let impediment s blocking their career progression remain for long.
Actal employees on the other hand are not treated like this because they are business as usual. As long as there is nobody yelling too loudly about those fires in the hallway they can be safely ignored.
The reward for hard work is more more work expected. Set realistic expectations and stick to it.
Also, we've been working longer hours post industrial revolution. Our farmer ancestors worked a lot less. Heck they even had siestas! So are you surprised?
I’m here before the bots wow
🚨🚨🚨 Bot account detected. No human can beat the bots.
@@pluto8404 Bite my shiny metal ass.
This article was basically an ad for Upwork that was written by AI. Agreed.
Tech needs to unionize to fight against these horrible implementations
It's to competitive. You can't have a union if there's someone who's willing to compete who isn't part of it. Companies will always hire people who aren't in the union over those who are.
GPT wrote me a script to invert PDF books so reading them won’t hurt my eyes.
It doesn’t work, obviously but it’s cute that it tried. And I have some cool new PDF CLI tools. No idea how to use them
Should have got it to write it Rust smh
I forget what terminal PDF reader I have (I think it's zathura?), but it has vim motions and color inversion as a motion. Anyways, I literally never use it, because I practically never read PDFs.
it's a myth that pre-industrial subsistence farming required working more hours
Farming for your family doesn't require that much time and effort. However, if you're farming for a civilization that consists of a significant population of non-farming citizens then it does require much time and effort and thus requires automation or economies of scale. Non-farming citizens are doctors, veterinarians, lawyers, teachers, builders, politicians, writers, artists, merchants, etc.
Plus subsistence farming is only super busy during sowing and harvest season, where so much preparation and processing needs to be done, of course also the occasional pest / natural disaster in the mix in worst case scenarios. The rest of the time it's just mostly watering and weeding.
In some locations subjects even have wine bread and cheese provided by the fiefdom, so full self sufficiency wasn't even required lol
That massively depends on what you're farming. Managing a herd of dairy cattle is a full time, seven days a week job. Even subsistence farming for just your family will involve vegetables, dairy, meat, etc. Preparing, sowing, maintaining, and harvesting crops. Feeding animals, collecting eggs, milking, and slaughtering and butchering animals. Anyone who thinks that's easy has probably never stepped foot on a farm.
What, how do you go from 99% of your population being farmers to almost nobody being farmers without those 99% being necessary for the job? Were the aristocrats really just letting all their serfs do almost nothing for all those millennia?
The scale of modern farming isn't what he's talking about. In the past people had a lot more free time than you realize. They worked hard when it was required, but it wasn't required half the time.
how does 'advanced clippy' make people think there's actually anything there?
On the topic of salary amount before or after taxes, in my country all advertised salaries are “after taxes”. As an employee, my employer pays my taxes, public healthcare and retirement/pension rate. I get a digital payslip each month so I know exactly how much they paid and I don’t have to worry about paying taxes. The state sends me a digital report each year telling me how much Ii’m expected to pay vs how much has already been paid, in case there is a discrepancy. So far I’ve only had to click the “I’ve checked and agree this is correct” button… once per year. Why the hell would I want to do the tax math/payment myself in 2024, when it’s all automated for me? I guess it’s an American thing or maybe something in the more developed west.
In the USA, the taxes you pay on income are not known at the time of earning it. They depend on your total income. It isn't a fixed percentage, but a progressive rate. And then there are deductions which reduce your taxable income, some of which you don't know what you'll have until the end of the year. You estimate how much you think you'll owe and your employer sends that estimated amount to the government. Depending on if your estimate is high or low you'll get a refund or have to owe taxes and penalties sometimes too. Some people also have to mail the government a check periodically to avoid being too far negative at the end of the year. Bottom line, your employer has no idea what your after-tax income will be.
Logistically, it has slowed me down. It adds pressure. Some do better, some do worse, but every body cracks eventually.
The thing with AI is it makes you more efficient. However, it makes unqualified people break things faster, amplifying their negative impact 10x
AI should only be used by people who:
A understand their job in deep depths
And
B know how to prompt an AI properly to get the best result.
that is true of a noob junior developer as well. If you use Ai on a project to learn it actually teaches you as you build as long as you try and understand what is going on and read up about other ways of doing it.
But someone once said of it works, it works.
I think also freelancers are hired when you need a project with their exact expertise, so they know a ton about that niche. If your normal employees don’t know as much about that area of development which is why you might hire a freelancer, they might not perform as well as the freelancer doing that one thing.
You have to a plan for who will take it over from the freelancer. The same problem that leads you to outsource it still exists when the project is done. The freelancer will go onto the next thing and you then have a project that you cannot support, extend or maintain. I saw this with one company I worked for where FPGA design was outsourced while internally no one knew FPGAs. We ended up with a ton of hacky workarounds because no one would touch it. It was actually a pretty simple FPGA design that anyone with a basic knowledge could have handled. But no one did.
@@username7763 yeah you need jack of all trades in these instances to bridge gaps. Which recruiters HATE
AI is a tool, period. At the moment, there is nothing but enormous cloud smoke and noise.
Massively underrated comment. Companies are trying to sell "AI-powered" products only because some sickeningly overenthusiastic, half-informed marketing exec read about it in one of the countless business advisor magazines (because those are always in touch with what the people want, right?), without any regard for the maturity and usability of any of it.
Nobody will be talking about AI 5 years from now… this AI talk has been happening for the past 60 years.
Btw, we don’t even have a definition for AI… it’s simply a corporate talking point.
AI has tried to happen, but we didn't have computers strong enough / efficient enough. Now we do.
@@tbunreall Define AI?
@@incawarrior5470black box pattern matching algorithms with parameters tweaked based on data
@incawarrior5470 define intelligence first.
@@hungrymusicwolf Exactly, then we throw the word artificial to it… then it becomes a term that can be used for any automated system.
LLMs were a success because the average population saw it as “wow, it is like us speaking”. After all, it is the condensed average.
This. Sad to say that even in this day and age, illiterate and poorly educated people exist despite the internet.
Now, imagine showing these people LLMs. They'd think it's Skynet made real.
Well they're outspooking the average. Its just we're figuring out the average or below average of humans all are in management. Its their jobs that would be most effective to replace. However in clown world its all the specialists being laid off.
We have been leveraging machine learning since before the AI explosion. The large language models are not all of AI, and expert systems are a completely different animal. It's true that employers will always squeeze more out of employees if they can, but we can have the computer do the tedious bits and let humans do what humans do better.
My thoughts on this:
"You? A lowly worker? Thought *you* would reap the benefits of this productivity increase? Oh my sweet summer child!"
It only works in our favor if we are using AI secretly.
I would like to point out the most (socially) competent people in charge are the ones who know they are bullshitting you non-stop, and also know everyone will smile and nod.
They also know ultimately if you just bullshit everyone and then somewhat let people do their things while paying them, solutions start appearing because the whole exercise is just social etiquette, yes, they know you speak ill about them at their backs, don't even think for a moment each part doesn't know exactly what the other part is thinking about the exchange.
Just to clarify, this isn't vranyo (and thanks to Putin's SMO for giving us this word), the difference between vranyo and bullshitting is with vranyo neither part expects anything to be accomplished (you pretend you work for us and we pretend we pay you), meanwhile western executives, while bullshitting you, still expect things to work, that's why western societies somehow are still stomping forward despite the manias and dysfunctionalities while the Global South needs to pay wumaos, bots and metodichkas to fool disaffected westerners into thinking there is a place they can flee from the bullshitting.
There is no place.
The problem with most people using AI is that unless you understand the requirements and can communicate them unambiguously, and then understand enough of the answer to verify it, then you are going to to hit the 'garbage in, garbage out' problem.
Yeah it doesn't help when they randomly go insane in hard to identity ways too.
I only use AI for HTML and CSS. I refuse to learn how to center a div!
Just and close tag at the end
I love how this thing talks about innovation increasing 2x. They empirically measured everything unprecedented. And guess what. It all turned out to be quantifiable with a single, convenient measurement. And it turned out we've got twice as much never-before-seenness in our whatsits. Great news.
Your point about job performance is SO spot on it's almost criminal. At my previous company I worked my ass of, I actually founded an entire new business area.
But then my dad got cancer and some other things happened and I "underperformed" that week when he was in the hospital (he also got blood poisoning) and voila, all of a sudden I got those "why are you not performing" bs comments.
Unbelievable.
Fortunately I found a new job and switched and it's currently great, but still. Really sad when you give so much to a company and they don't give a shit.
Yes they don't care, and upward movement is a lie, brown nosing is the only skill needed to go upwards, easier to move to a better job in another company. Also your just wasting your life working, if one has no debts then why do we need to work.
As an artist, I have to thank artificial intelligence. It has increased my work and there are many people who want to learn drawing after using artificial intelligence XD
Really that makes me actually smile! I was hoping it would inspire more people to get into art!
that's why i don't fear technology.
if it makes things better, now you have more work to implement the new thing
or
if it solves a problem, then now some things that were not possible before are possible now... and the end is more work
some jobs are lost... but the amount of work always goes up, it will never stop going up
More jobs are replaced with fewer jobs at any given company using AI to increase worker productivity
Pipedream: technology will decrease workloads and give people more free time!
Reality: Here's 500 more things to work on... and yes, if this looks like the existing work other teams would've worked on, that's because it actually is. NO 15 HOUR WORK WEEK FOR YOU! Good luck!
**Looks around and realizes that no other team exists in the office.**
_New Objective: SURVIVE_
Management is an evil position to hold at this point
Dealing with the garbage output of AI is such an unbelievable headache, and yet it's freaking everywhere.
Especially for looking for basic things like stock photos or 3d renderings
This improved my wellbeing
That's because right now, AI is NOT being used to make their jobs easier.. It's being used to help collect further data on how their jobs are done, so better AI can be trained to fully replace them.
Oxford comma mentioned, let's gooooo
Woohoo!
There will never be a time when workload will be reduced
So we increase productivity not to work less and focus on technology breakthrought but to produce more stuff people don't need 😜
It seriously could have allowed people to work less sadly that requires competent management.
this guy is just reading articles to me, why am I enjoying this
Freelancers don't do maintenance
Freelancers are better perceived performance because full time employees are asked for everything they can give the employer, and freelancers have billed hours. They pay, they get. Full-time workers are used to the business and know the company but that is not valued at all. They are thrown under the bus all the time.
Prime believes in meritocracy in the c-suites 🤣😂🤣
What I'm hearing is from this article is "your employees are burned out by your AI-driven expectations, therefore you must hire some highly trained free lancers that will set up a castle of AI-native processes, and then close the contract as soon as your expectations are no more economically advantageous to them, leaving you with even more unhappy employees".
They should ask the solo developers and small teams how AI affects them. People who don't have managers that increased expectations, who personally chose to include AI in their workflow, and designed their pipeline around it.
For personal projects it greatly enabled me to learn and complete them
I work 40hrs for my salary. If I somehow need to work more, I "charge" the standard overtime rate via the time reporting system.
Wages and salaries do not work the same all over the world... :)
What? Don’t you know that computers are smarter than humans?
C-Suit & Investors want workers to do more so they can all do less (and get rich while doing it).
Flip actually adding the sparkles was so wholesome. ✨
"that would also be known as react" had me ROLLIN'.
I know a guy who had 3 contract jobs at once and anytime he was drowning in work he just sub contracted someone from India, he did say that didn't happen that often.
Its called hallucinating. Its when data is pulled from dissimilar subjects relating to similar keywords. Ai cannot verfiy its own information and spits out bs chatbot garbage. Most actual people do this too, its not just ai. This world needs to recognize the limits of stupid low iq peopleand that they will never be productive, just robotic.
Remember a while back my comment "AI = Productivity goes up, Quality goes down". Well I was wrong. It looks like I have to update:
AI = Productivity goes down, Quality stays down
Summary:
A study shows that 77% of employees report that AI has increased their workload and hampered productivity, while 96% of executives expect AI to boost productivity. There is a disconnect between the high expectations of executives and the reality faced by employees, who feel overwhelmed and are experiencing burnout. The discussion around the use of freelancers is seen as potentially biased, as they may be more productive but often provide short-term solutions. To fully leverage the benefits of AI, companies need to fundamentally rethink their work models and integrate alternative talent pools. Overall, it is emphasized that the integration of AI into the workplace presents both opportunities and challenges, making it crucial to develop a realistic understanding of its impacts.
AI will only be as good as the person programming it. Prove me wrong
Yup
Proving you right is the hard bit. Function approximation isn't intelligence, current "intelligence" is indeed artificial
If you think about it, using AI is like being a coach for a team of both geniuses and complete knuckleheads. Your main job? Yelling at them to get the simplest tasks done!
WTF FLIP??? What is that thumbnail KEKW
4:56 this is so true at companies. I’ve made this mistake before. You start off never taking PTO and skipping your lunch. It becomes harder to reel it back afterwards. Precedents are important to set
To clarify, AI didnt fade out and come back in the last 50 years due to algorithms, but due to compute. in the 1970s we didnt have enough compute. Right now we are slowly getting close to the estimated compute of the brain, back then this was exponentially far away. To think there will be another AI winter because of the last AI winter just shows poor understanding of the subject.
And compute is increasing on a logarithmic curve, so what does that say about my understanding of the subject?
@@mage3690 "logarithmic curve" is actually very non-descriptive. If you understand that all you need for intelligence is scaling up the model components, so either architecture, number of parameters or number of quality tokens, you basically get more than most people including prime himself. Scale is all you need. If you increase compute you allow for more parameters, longer training aka more quality tokens and through research, the only part prime knows, you get minor improvements to the efficiency. But nothing you cant get with more compute.
We are not getting close to the estimated compute power of the brain, it's not even close.
@@saharatul estimated compute is what, 10^25 or 26? We are at around 10^24ish.
@@pottedrosepetal6906 There are like 100 trillion synapses in a human brain. the latest Nvidia B100 has a measly 100 billion transistors
Hey Primetime, hope you see this. One thing you have to consider about freelance or temp workers compared to full time staff is the implied job security. Most perm workers live under the assumption that they have their job and all the benefits/safety nets it comes with, like consistent health insurance included, labor protections (I.e. unemployment, COBRA, etc.). freelancers don’t have that, so they’re able to demand a higher wage to make it worth the time, loss of those benefits, downtime between freelance contracts, and relative job uncertainty
I just wanted chatgpt to do some boring 20 lines copy paste pandas code, first suggestion, revert the entier dataset, do calculations and revert it back. Suggestion two, only revert it once and have two copies, suggestion three, just use shift (with a wrong amount), and halusinating it would work with an output example. Ye, it took longer than just writing the code myself (heck, opening chatgpt and asking probably took longer), and it didn't even end up giving one good solution. And I rarely ever try anymore, the solutions are trash, and most things doesn't work. Its quite alright for very short basic problems, but its also not something most programmers need much help with.
Point #4 basically says instead of planning projects as a set of specific jobs that you need done to get the project done (I need someone to get me coffee every MWF at 10:00 AM) to skill-based (I need someone who can walk fast and handle hot substances). To add AI to the mix that would be an employee that can program a VoIP AI to call you around 10:00 AM every Monday Wednesday and Friday asking you how you'd like your coffee, take your input, and map your input to an API call that orders a coffee with your preferences and have someone deliver it to you. Insane and stupid, but scalable.
5 seconds is crazy
5 seconds is indeed crazy brother 🔥🔥
You can tell which RUclipsrs had good reasoning skills by those that didn't need two years to arrive at this conclusion.
Problem is not all AI is the same. Claude 3.5 is producing very good code. Google Gemini? Not so much. Python, Java, or C++, HTML, JS, they are very good at. Others, seem to be very hit and miss. I expect this to only get better rapidly judging on how much better the last year has given us.
Gemini makes a lot of mistakes overall. Can't really rely on it at this point.
We started paying for Copilot & Chat GPT around January of 2023.
A large chunk of our current work is completely refactoring work created with the assistance of machine learning tools.
Most of the markup is fine, map/reduce/filter/sort style functions are fine, but anything that required implementing a solution vs autocompleting a formula has ended up with tightly coupled code, that is very slow for large operations.
The ML stuff is fine for small & narrowly defined functionality, but it does not work at all for enterprise level software in the long-term (at least not so far)
We still pay for ML services, but frankly, I'm spending more time reviewing and rejecting code as a lead than ever.
On one hand, I review commits a lot earlier in the process than I used to, to avoid bad code (I'm not sure, is that actualy a good thing or a waste of time?). On the other hand there's so many more "high level" issues, that people are continously re-doing stuff.
Frankly, the box has been opened. Devs (on average) will allways take the easy way out, which now means using ML code without thinking it ober. Quality control is going to become the big differentiator in the future. I don't know If that's sustainable, I'm personally mentally exhausted.
Come on it’s not that good. All these LLMs are only useful for completing trivial tasks. It definitely makes those tasks faster though.
@@nuvotion-live Define "not that good" ?
@@CommanderRiker0 doesn’t live up to the hype. It’s still useful but way overestimated
I don't work in tech, but I work as an artist in gamedev this holds up. One project started implementing ai in the pipeline, the amount of artists working on this project was significantly decreased yet deadlines became tougher (like a task that beforehand was estimated to last 40h now was estimated at 16h) Artists who had to work on that project complained a lot, because although ai did make the process faster, but not by that much, there were still plenty of mistakes that had to be fixed after the ai and some tasks it did downright terribly. Artist were very overworked for the same pay as before.
This is a textbook advertorial. Which is just an ad disguised as an article.
To the freelancer segment, I would also add 2 things :
- highly paid free lancers are hired for their expertise, full time employees not always, you might be hired to code but have to do powerpoints instead
- freelancers participate to the estimations because their business depends on it. In some companies, Johnny the manager who has not coded in 6 years does the estimations for bob the coder, and then bob does not meet the deadline
I've worked jobs with exact opposites in the US and UK, and the US side were paid less, got less leave, and called their bosses "sir" much more.
Can we start a union of programmers that only work to be paid with a gold-backed currency?
AI is even affecting basic retail worker's jobs. They set it to organize the tags on the shelves, and it doesn't even understand counting. There is a system and there are sections and levels to the numbering system, but now they'll be in the wrong order, in the wrong sequence, going backwards, and organized so that products dont fit in the first place
India is the most common place for outsourced remote work for medical billing (and we've seen them used for "AI"). Malaysia is also common.
They oft have Morning/Night shifts too, so timezones aren't a major issue for them. Instead, their ridiculous amount of holidays and minimal competent staff is more likely the issue.
(edit: their more competent staff doesn't usually do the remote/cheap labor stuff. it's just a numbers thing)
(also, the whole "before or after taxes" was pertinent for your question. places like the UK deduct and file the taxes for you, so you really only care what you make after tax)
the thing about subsistence farming is you can only do so much before you just have to wait for things to grow, so no, people who were subsistence farming were not working all day every day. and they stopped working when the sun went down, electric lighting increased the average workday significantly