back to article Study shock! AI hinders productivity and makes working worse

Bosses expect artificial intelligence software to improve productivity, but workers say the tool does the opposite, according to a survey by find-a-workplace research org the Upwork Research Institute, a limb of talent-finding platform Upwork. The survey elicited responses from 2,500 workers across the US, UK, Australia, and …

  1. Joe W Silver badge

    Ah, the old one...

    as the BOFH remarked when the Boss asked for something:

    "and you want us to make it 'better' using 'technology', and if we just 'thought out of the box' the solution would 'stare us in the eyes'"

    Or something like that. Didn't the boss get cut by the PFY with a linoleum knife then?

    Remember, folks, if you cannot read ElReg and the BOFH and declare it as "essential training" you have the wrong job (that said, I do work enough, and currently my code's compiling / my SQL query is stumbling along)

    1. Doctor Syntax Silver badge

      Re: Ah, the old one...

      "my SQL query is stumbling along"

      A well-crafted SQL query can stumble along for days. it's just a matter of knowing which columns aren't indexed.

      1. Helcat Silver badge

        Re: Ah, the old one...

        "it's just a matter of knowing which columns aren't indexed."

        Don't...

        I've been asked to look into why a particular job takes ages to run. Oh, they'd checked for missing indexes (I'd shown them how to do that, at least) and there weren't any.

        Only the query was referencing tables in other DB's (without using synonyms) so the index hints don't work.

        So it was no surprise to find there were NO indexes on any of the tables. Not a one...

        1. sedregj Bronze badge
          Windows

          Re: Ah, the old one...

          ... and don't forget the power of the prime numbers for scheduled tasks.

          I nearly always use prime numbers for number of mins past the hour or since boot or whatever, when scheduling tasks. It helps avoid collisions.

          Two jobs: one at five and one at 10 periods. They will run at exactly the same time, every 10 periods. Change one to seven and the other to 11. Now they will only collide once per 7x11=77 periods. Handy.

          Now extrapolate to three jobs: 5, 10, 15 and change to 7, 11, 13 (or 17).

          Many schedulers will also allow you to inject a random spread too and combined with prime numbers gets you a great way to spread the load for regularly scheduled tasks.

          1. Eclectic Man Silver badge
            Thumb Up

            Re: Ah, the old one...

            "Now extrapolate to three jobs: 5, 10, 15 and change to 7, 11, 13 (or 17)"

            That is really neat! (Have an upvote on me.)

            One interesting (to a nerdy mathematician) is that 7x11x13 = 1001. Which means that the test for divisibility by 11 that you learn in primary school (sum alternate digits, take the difference and if it is divisible by 11, then the whole number is) can be used, base 1000, to test for divisibility by 7, 11 or 13, rather than using cumbersome or less efficient code. This was actually useful for me in some number processing I did recently.

            Also, I recall that Sir Paul Nurse was interested in cell division, and instead of looking at the cells a regular intervals he looked at irregular intervals, found a fundamental mechanism of cell division and got the paper (and a Nobel prize for medicine).

            1. Doctor Syntax Silver badge

              Re: Ah, the old one...

              I've used the same logic to set boot intervals for running fsck on different partitions.

          2. Sparsely the Lion

            Re: Ah, the old one...

            > I nearly always use prime numbers for number of mins past the hour or since boot or whatever, when scheduling tasks. It helps avoid collisions.

            > Two jobs: one at five and one at 10 periods. They will run at exactly the same time, every 10 periods. Change one to seven and the other to 11. Now they will only collide once per 7x11=77 periods. Handy.

            It's a good point that I don't wish to distract from but the pedant in me feels obliged to point out that 5 is a prime number. :-)

            1. M.V. Lipvig Silver badge

              Re: Ah, the old one...

              Yes, but 10 is not.

              1. Sceptic Tank Silver badge
                Angel

                Re: Ah, the old one...

                Depends on which base you're using.

                (10 types of people: those who understand binary and those that don't, etc.)

              2. Michael Wojcik Silver badge

                Re: Ah, the old one...

                Yes, but the point was that it wasn't necessary to change the 5 to have two prime intervals (and thus a maximal LCM for the intervals). Of course using two primes that were both greater than the original value they replaced gives a larger overall LCM, but if there was some reason to have the task that ran at interval 5 run as close to its original interval as possible, there's no need to change it to 7.

                I've used prime values for things like pseudorandom parallel testing (select at (pseudo-) random from a list of primes in the appropriate range) of distributed systems for a similar reason — exercise a wider distribution of timing effects, to try to catch races. (Of course with those sorts of systems you do also want to test collisions, so it's good to have multiple runs with different distributions.)

          3. Clarecats

            Re: Ah, the old one...

            "Some periodical cicadas emerge every 13 years and others emerge every 17 years."

            https://naturalhistory.si.edu/education/teaching-resources/life-science/periodical-cicadas

            This is a mechanism to disrupt predator cycles.

  2. Anonymous Coward
    Anonymous Coward

    C-Suite are often divorced from reality. They see these tools as golden tickets to being able to hire low-skilled labour at reduced cost as opposed to just tools that can aid in performing a task. The real problem comes when there is a failure and you don't have any skilled engineers left who can debug and resolve the issue. For example, I've sat in a couple of technical interviews where I've asked simple questions only to be met with the answer "Well I run this automation script....", and the candidate was completely unaware of how to do the job otherwise.

    1. hoola Silver badge

      That is why companies target C-Suites with all the sales bull. The will buy anything if the marketing looks good.

      How else did CrowdStrike get to where it is?

      Why is everything now marketed with some AI elements?

      It is what o er paid boses want to hear. Anything that is not:

      Cloud native

      AI driven

      Subscription priced

      Is considered old fashioned or the term I detest "legacy ".

      1. Triggerfish

        I find it sort of depressing how easily the marketing and sales people full for the exact same bullshit they craft themselves.

        1. vtcodger Silver badge

          Hmmm. Perhaps we should start a rumor that 900% savings can be achieved by replacing the Marketing and Sales staff with AI agents operated by prison labor paid $0.25/hr. Let's see how well that one flies.

          1. spacecadet66 Bronze badge

            If any business function can be replaced with stochastic parrots, it's marketing.

        2. Doctor Syntax Silver badge

          Depressing but inevitable given that they believe it. To do otherwise would invite too much mental dissonance.

    2. Anonymous Coward
      Anonymous Coward

      >>>C-Suite are often divorced from reality.

      Well, it is full of C.s.

      1. ecofeco Silver badge

        BA DUMP BA!

        Well played!

      2. spacecadet66 Bronze badge

        I thought that was "C"s as in "mediocre students" until I remembered, this site has a large British readership.

    3. Jedit Silver badge
      Pirate

      "C-Suite are often divorced from reality."

      And divorced in general.

      The fastest way to increase productivity and profitability in a company is to fire the entire C-section. Unfortunately, it's the C-section that makes those decisions. And so we carry on.

      1. Blank Reg

        Re: "C-Suite are often divorced from reality."

        And with the currently bloated c-suite salaries getting rid of them would result in quite a cost savings without impacting productivity

      2. Anonymous Coward
        Anonymous Coward

        Re: "C-Suite are often divorced from reality."

        "C-section"

        Too posh to push?

        1. Jedit Silver badge
          Trollface

          Re: "C-Suite are often divorced from reality."

          As I said the last time someone enquired about that usage: it's because the only way to get anything out of those c**ts is with a sharp knife.

  3. Anonymous Coward
    Anonymous Coward

    AI seems to encourage magical thinking

    That pretty much sums it up: The upper levels of most companies run on magical thinking, regardless of technology.

  4. Moldskred

    "Coincidentally, the Institute – part of the Upwork platform for hiring freelance workers – suggests that hiring freelance workers can help."

    Ah! I was _wondering_ why the survey had been designed so as to poll double the number of freelancers as full-time workers.

  5. b0llchit Silver badge
    FAIL

    Wrong productivity trajectory

    AI is supposed to replace the workforce! You don't keep them on board and pay them. You replace and fire the lazy meatbags.

    Hey, C-suite, get going doing your jobs. You need to start firing all those meatbags and cash your well-deserved hundredfold bonus in the process!

    /s

    1. ariels-again

      Re: Wrong productivity trajectory

      Exactly! And my research shows that AI is best poised to replace the C-suite, where it can leverage its inherently equal knowledge across the board to deliver equivalent performance at lower cost and latencies. Oh, and replacing all those expensive workers means we can afford the increased ai spend.

  6. lglethal Silver badge
    Facepalm

    Wow, what Insights...

    Management divorced from reality! Workers Overworked! AI is not actually helpful! Water is wet! Bears shit in the woods! The Pope is Catholic!

    Amazing insights in this report... Truly wonderous! Can I get paid to produce such reports too?

    1. daflibble

      Re: Wow, what Insights...

      No that's AI work these days

    2. SundogUK Silver badge

      Re: Wow, what Insights...

      Not sure about the Pope, to be honest.

      1. Sceptic Tank Silver badge
        Big Brother

        Re: Wow, what Insights...

        I can immediately think of two types of water that isn't wet.

        1. Michael Wojcik Silver badge

          Re: Wow, what Insights...

          Three, maybe; is a hydrogen-oxygen plasma "water"?

          (And could we count each of the different phases of ice separately? Wikipedia says currently 21 are known. It's not all common-or-garden ice,1 you know!)

          1Also known as "vanilla ice", not to be confused with "Vanilla Ice", which while briefly common is now fortunately quite rare; or "vanilla shaved ice", actually a compound found mainly in Hawaii and Italy.

      2. Anonymous Coward
        Anonymous Coward

        Re: Wow, what Insights...

        Nope...

        https://m.youtube.com/watch?v=RMf4OtC7SXY

        (looks a bit like Stephen Fry)

    3. Michael Wojcik Silver badge

      Re: Wow, what Insights...

      The point of a study like this isn't its conclusions, but the data it amasses to support them.

      Anyone can claim that gen-AI decreases productivity, just as anyone can claim it increases it. And anyone can drum up some anecdotes and sophomoric arguments to support either position. (There are also sophisticated arguments, but most people don't want to go to the trouble of understanding them.) What counts is data (modulo methodology problems, scope, etc).

  7. daflibble

    same as it ever was...

    Yep another technology to use as a tool. Problems are not so much the tools or technology, it's knowing what you are trying to do with them, then building the processes.

    1. Pascal Monett Silver badge

      Re: same as it ever was...

      Not quite : it's another beta version of software used in production.

      It was ever thus.

      1. Michael Wojcik Silver badge

        Re: same as it ever was...

        Given the state of explicability and alignment research, it's unlikely autoregressive LLMs will ever come out of beta. I think we need another term for this one; "beta" implies a smidgen of hope.

  8. Bebu
    IT Angle

    "When compared with full-time employees, more freelancers claim to be AI-ready"

    MRDA

  9. Anonymous Coward
    Anonymous Coward

    What is rather ironic is that the jobs most suited to being shunted over to AI are the C-Suite jobs.

    Think how much money would be saved there!

    I think this skit says it all....

    https://www.youtube.com/watch?v=WCZt4mCcCV8

  10. Denarius Silver badge

    that explains much

    so LLM produce something without context. When put into a loop gibberish results. Would this apply to "real" intelligence also ? So does this explain the stupidites emitted from manglement , much academia and political classes in their bubbles ?

    1. TheMaskedMan Silver badge

      Re: that explains much

      "So does this explain the stupidites emitted from manglement , much academia and political classes in their bubbles ?"

      Of course. See flat earthers, creationists etc. all pretty much exclusively consume their own bullshit, all are incoherent. Garbage in, garbage out.

    2. Antron Argaiv Silver badge
      Thumb Up

      Re: that explains much

      I have yet to be convinced that "AI" (e.g. LLMs) is anything more than a party trick. Grab a corpus of words and phrases, extrapolate the probabilities of adjacency, then mix and output? I'm not knowledgeable about the technology, but everything I see seems to indicate that any "intelligence" in the process is probabilistic or Markov-chain driven.

      Therefore, asking "AI" to produce anything both novel AND useful (as in a business) is a pipe dream...you're only going to get what was in the initial input corpus, ground up and regurgitated.

      It may appear novel, but when one looks deeper, there's no thought behind it.

      Happy to be educated to the contrary

      1. JRStern

        Re: that explains much

        >Happy to be educated to the contrary

        So would everyone involved in any way with LLMs, LOL.

        >It may appear novel, but ...

        No, novel is just what it's not even supposed to appear. It's purely retrieval. Perhaps it can retrieve intelligently but if you give it a truly novel situation there will be nothing available. You then get one of these lengthy responses listing all the possible associations - but in many cases you already knew all that and it's only tedious to have it thrown back at you.

        In many ways a four-function calculator, or an Excel spreadsheet, outperforms any LLM at today's level of technology.

        Check back every five years for the next century and I'll bet you see progress being made, LOL.

      2. Eclectic Man Silver badge
        Boffin

        Re: that explains much

        Well, there has been some improvement in the solving of mathematical problems. In the mathematical olympiad an AI solved one of the geometry problems in a few seconds:

        "A pair of new systems, called AlphaProof and AlphaGeometry 2, worked together to tackle questions from the International Mathematical Olympiad, a global maths competition for secondary-school students that has been running since 1959.

        ...

        Unlike a human mathematician, the systems were either flawless or hopeless.

        ...

        The other system, AlphaGeometry 2, similarly pairs a language model with a more mathematically inclined approach. But its success at the narrower field of geometry problems was startling: it solved its problem in just 16 seconds.

        ...

        The 'easy problem:

        Let ABC be a triangle with AB < AC < BC. Let the incentre and incircle of triangle ABC be I and ω, respectively. Let X be the point on line BC different from C such that the line through X parallel to AC is tangent to ω. Similarly, let Y be the point on line BC different from B such that the line through Y parallel to AB is tangent to ω. Let AI intersect the circumcircle of triangle ABC again at P ≠ A. Let K and L be the midpoints of AC and AB, respectively.

        Prove that ∠KIL + ∠YPX = 180◦."

        From: https://www.theguardian.com/technology/article/2024/jul/25/google-deepmind-takes-step-closer-to-cracking-top-level-maths

      3. Michael Wojcik Silver badge

        Re: that explains much

        Happy to be educated to the contrary

        I am very much not a fan of LLMs or Deep Learning in general,1 but I recommend learning something about the technology. Reducing it to a Markov process doesn't provide any useful insights, any more than reducing it to arithmetic or Boolean logic does.2

        Hackaday a blog post by Grindberg if you want a non-mathematical explanation. Personally, I like this (fairly long) piece by Wolfram.

        If you want to dig deeper, a lot of papers on LLM research get relatively detailed summaries posted to Lesswrong. Lesswrong is of course (in)famous as the home, more or less, of a prominent contingent of notkilleverythingist opposition to AI, a position you might have some opinion about; but there's plenty of useful and interesting material there regardless of your stance on AI Gonna Kill Us All.

        1What about diffusion models? There I think the primary issues are social; I don't really care that we don't have good analytical tools for diffusion models, or that they make mistakes, etc. They're mildly interesting from a CS perspective.

        2Any computable function (probably) can be represented as a Markov process. So what? Sometimes that's a useful thing to do, but often it's just a truism that hasn't told you anything new.

  11. Michael Hoffmann Silver badge
    Unhappy

    Sometimes, in my idle moments, I wonder what would have to change and how you would set up economies and businesses where the turds don't float to the top.

    The system is broken, that's pretty clear, but what checks and balances would you need - and how would you implement them past the people who have greatest interest and influence of things NOT changing, even if it's approaching collapse at alarming speeds?

    Haven't we tried just about tried everything and the result is always gravitation towards some variation of "sociopaths kissing upwards and kicking downwards"?

    1. michaelaubert

      defeatism will unfortunately not help

      I feel plenty of it too but we will only be truly F*ed when we give up trying.

    2. ecofeco Silver badge
      Holmes

      You and every great thinker and philosopher in the history of mankind for the last ten thousands years of recorded history.

      And the answer is: it's never going to happen. The best, the absolute best that can be done is to lessen the effects of the smell. And even that is only temporary.

    3. Eclectic Man Silver badge
      Flame

      re: 'The system is broken'

      Sorry but I have to disagree. The system is NOT broken - IT NEVER WORKED IN THE FIRST PLACE.

      The use of the phrase 'the system is broken' implies there was a time when it did work, everyone was (reasonably) happy and the basis can be kept and with a few 'tweaks' it can be 'returned' to a satisfactory system. This is wrong in many cases. What we have is a system that allows self-centered idiots with high confidence, entitlement and personal connections to dominate the upper echelons of business and management (and elsewhere - look at the proportion of privately educated UK Olympic competitors (33%) compared to the 7% of the UK population who go to private schools*)

      That system was designed to favour the rich and well-connected, and is working very well indeed (for them)**.

      But have an upvote for the sentiment.

      * https://www.theguardian.com/education/article/2024/jul/24/third-of-team-gb-athletes-for-paris-olympics-educated-privately

      ** I'm NOT bitter, not at all, I always look like this.

    4. Doctor Syntax Silver badge

      In any hierarchical organisation those at the top of the hierarchy the only ability you can be sure will be found in those at the top will be the ability to climb hierarchies*. Any other abilities will be a bonus.

      * Exception has to be made for those where the position is inherited. There what you get is a matter of pot-luck, hence the saying "rags to rags in three generations" is so often true.

      1. Michael Wojcik Silver badge

        From an evolutionary point of view, the opportunity to inherit power1 counts as an "ability" to ascend the hierarchy too. You can view it as an illiquid, non-fungible, mostly-non-transferable2 capital good.

        So you don't really need your exception.

        1Which includes access to resources, i.e. wealth.

        2Inheritance is transferable via social mechanisms such as disinheritance, adoption, marriage, etc. For historical and legal reasons societies typically impose a high degree of friction on such transfers.

  12. Pete 2 Silver badge

    Take this job and ....

    > one in three employees say they will likely quit their jobs in the next six months

    But we all know that they won't.

    1. michaelaubert

      except for the ones who totally will quit

      It took a whole year since the insulting RTO email but I found a better role and am currently serving my notice period.

      Apatheism isn't particularly helpful here.

    2. ecofeco Silver badge
      Coat

      Re: Take this job and ....

      Have we forgotten the ones who will be volun-told to leave?

      Contrary to whatever plans they have.

  13. Doctor Syntax Silver badge

    "AI seems to encourage magical thinking."

    I think it's more along the lines of being in top management is associated with magical thinking. I'm not too sure about which way the causality runs.

    1. ecofeco Silver badge

      GIGO, from top to bottom.

    2. David Hicklin Bronze badge

      "AI"...the new AGILE will fix everything that is wrong....

  14. Howard Sway Silver badge

    they're spending more time reviewing or moderating AI-generated content

    This is a mistake. If you're forced into using AI to produce stuff, you should just go with whatever it produces. The boss heard it makes you more productive, so just use it to produce more stuff. At some point complaints will start to arrive that you're producing large amounts of irrelevant nonsense, but you can point out that it wasn't you that produced it : it was the AI you were ordered to use. At this point you'll be told to start reviewing the output, so you can ask the AI to review it, as the AI is still mandated for work.

    Presumably at some point your exasperated boss will shout "Stop using the AI to do everything!".

    1. Missing Semicolon Silver badge

      Re: they're spending more time reviewing or moderating AI-generated content

      Problem here is that the staff will still be held responsible if the AI product is faulty. "I thought you were checking this?"

  15. Nematode

    Never mind AI. We need to something first about Natural Stupidity.

  16. Handlebars

    where's the detail?

    What does highly skilled at using AI mean? I followed the link and still no idea. Is it 'i can use chat gpt to get recipe ideas' or 'my computer vision project will be scoring the gymnastics at the Olympics'?

    1. Irongut Silver badge

      Re: where's the detail?

      Based on the people I've worked with over the years who claimed to know how to use a computer it means they have never used AI, have no idea what an AI is or how it works and will blame the IT dept when they can't get their AI to work. They will probably also be paid more than IT.

      1. ecofeco Silver badge

        Re: where's the detail?

        That would be a good bet.

    2. Michael Wojcik Silver badge

      Re: where's the detail?

      Prompt engineering, mostly, particularly for creating specialized system prompts for custom chat applications. Think what you may of this (and I'm against the whole application myself), this 1) is a skill (it requires education, practice, and continual improvement) and 2) is in demand.

  17. ChrisElvidge Bronze badge

    Am I missing something here?

    "Half of respondents were C-suite execs, a quarter worked full time and the remained were freelancers."

    So three-quarters of the responders didn't work full time?

    1. ecofeco Silver badge
      Mushroom

      Re: Am I missing something here?

      How much do you know about CxOs?

      They rarely work full time. They often sit on more than just one board. And rarely full time on any of them. Many have taken their golden parachutes and are just hanging out looking for the next company to fleece. Most likely one recommended by their peers at the country club or the right school tie.

      None of that is hyperbole. I've walked those halls and been the fly on wall at one time or another. Not to mention daily reports hear and in the news of the "our betters" constantly committing crimes. They are, to a person, lazy, useless, psychopaths and sociopaths. The one in a million exception is just that. An exception.

      And the shite ALWAYS rolls downhill.

  18. heyrick Silver badge

    In the boardroom, AI seems to encourage magical thinking.

    There. Fixed that for you.

    1. Doctor Syntax Silver badge

      No encouragement is needed.

  19. Boris the Cockroach Silver badge
    Boffin

    I'm still waiting

    to find out how an 'AI' will help me find and replace the faulty encoder on a machine ... especially when the only error I'm getting is "Sync failure"

    The fact that I know which one it is (mostly due to me unplugging it at 7.45am before the boss got in) is irrelevent (also fault finding the machine and replacing the problem takes 2&1/2hrs during which time I should be in the end of month production meeting........) but in the case of a real failure..... how is AI gonna help me? (besides being used to generate a list of excuses)

    1. Michael Wojcik Silver badge

      Re: I'm still waiting

      Hey, something has to suggest turning it off then on again.

  20. neilg

    Management drank the Kool Aid

    Forgive me, I'm from the UK, I've googled Kool aid. Why would anyone of sound mind drink it in the first place?

    1. Doctor Syntax Silver badge

      Re: Management drank the Kool Aid

      Who said anything about sound mind?

    2. Michael Wojcik Silver badge

      Re: Management drank the Kool Aid

      Poe's Law applies here — I honestly don't know whether you get the historical reference.

      (Aside: The "drank the Kool-Aid" idiom is really quite disrespectful to the victims, and a good writer ought to eschew it in most circumstances.)

  21. DS999 Silver badge

    I think AI is overhyped

    But it is way too early to determine the long term effect on productivity. The amount of effort to integrate AI into workflows will swamp any benefit initially, especially in companies starry eyed enough about AI to be able to participate in surveys like this one.

    Just look at one place where AI can help, front line customer support. Replacing IVR phone trees with something able to carry on a normal conversation will improve the experience for customers and reduce the number of actual human support staff, but the AI will have to be trained on the material, you have to insure it won't go rogue, you have to do pilots to judge customer reaction and fix issues that identifies, and so forth. It isn't like the industry is anywhere near mature enough today that you can simply buy an "AI server", plug it into your network, and pull the plug on your IVR.

    1. ecofeco Silver badge

      Re: I think AI is overhyped

      LOL! Frontline support is the LAST place AI will be of any use.

      It won't stop the CxO wankers from trying, but it will fail.

      If you've never worked frontline, there is no way to explain the gigantic volume of cockwomble, numpty pillock wankers you have to deal with. And your hands are tied by the most Kakfaesque rules and polices from manglement, design to prevent quick solutions.

      AI will lock-up in a feedback race condition.

      Did you think BOFH was fiction? Only the poetic justice part of it is fiction.

      1. Michael Wojcik Silver badge

        Re: I think AI is overhyped

        The beauty of it is that customers are already using gen-AI to create their queries to support. (There are extensive discussions of using it to craft letters of complaint, for example, or to generate the questions for website chatbots.) So we have two resource-intensive, misaligned, often-hallucinating automated processes winding each other up, often with no useful outcome.

        We're starting to see the same thing with email and other forms of communication. It's already possible to do it (in practice, using commercial services on both ends) with voice calls as well.

        Soon humans will be freed from the bother of participating in communication at all.

    2. tsprad

      Re: I think AI is overhyped

      Yeah, customers just LOVE those IVR phone trees. And hold music. /s

  22. Anonymous Coward
    Anonymous Coward

    Don't get me started

    I hate marketing wank, truely hate it. I sometimes feel "Am I just moaning for no reason" then our old, now sadly left boss joined and agreed with me. We never fell for marketing bollocks, it was great. Avoid Dark Trace cause of it and shortly after they had a scandal.

    Now, sadly, people have been in charge for over a year with no fucking clue. So all fall for the marketing wank. In our weekly meeting with an external partner I would point out "Moving fully to the cloud is going to be more expensive". Because I don't kiss arse in the office, because I just keep to myself, no one listens to me so I stopped saying it. And the "partner" kept jumping in and claiming no it won't. So they listened to them.

    Role on months later and management have finally been given then bill and asking why its not saving money and is more expensive. The partner has finally admitted it will be more expensive. Everyone ignores I told them that on and off months earlier.

    We are currently testing copilot with management using it to do their job for them but staying on same wage. I've said time and again, if not careful with it, it will cause a Horzion scandal.

    Already has at DWP so they had to stop using it or double check its work.

    Post Office have yet again fucked up with their "automated" system for checking stamp fraud that has, like the Horizon scandal, been charing people for arrived letters with "fake stamps" that weren't fake. So people had to pay for their post.

    And finally we have copilot recording our meetings and then summerisies them. I know full well people don't bother to check the recording matches. Our last meeting it said I had said stuff that was never said in the recording. Those types of issues will be used years later in potential legal actions "Well Copilot said you said this in the meeting so you must have. So why did you miss those details and all those people died?" And you argue till yoir blue in the face that no, Copilot is wrong I never said that. "Copilot is never wrong and you're the only one this has happened to".

    In a way, much like most, I'd love to retire early. Not much more I can do in other jobs. I still like IT but sick of people not listening and sick of the Peter Principle and increasingly sick of the cloud and subscription only fees instead of being able to buy the software on DVD. But it can never happen. Stupidly started work late so will have to continue till I die.

  23. nobody who matters

    As soon as I hear someone referring to what we currently have as 'AI'. I know immediately that I am dealing with an idiot!

    It <isn't> intelligent, and the only thing about it that is artificial is the belief that it <is>.

  24. Dr Sendy

    C-AI

    Let me re-frame that question for you.

    Why should we not be getting AI to do the C-Level's jobs. At least the AI would be aware of the limitations of technology.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like