Advertisement
Help Keep Boards Alive. Support us by going ad free today. See here: https://subscriptions.boards.ie/.
https://www.boards.ie/group/1878-subscribers-forum

Private Group for paid up members of Boards.ie. Join the club.
Hi all, please see this major site announcement: https://www.boards.ie/discussion/2058427594/boards-ie-2026

Will AI take your job?

2456716

Comments

  • Posts: 0 [Deleted User]


    turns to sh*t? Paypal, Tesla, X, Space X? Starlink? The richest guy in the world?

    He can probably afford all the child support for every kid in America and every drug from Columbia.

    Same argument as football fans saying 'this player is sh*t….that player is sh*t' after a loss.

    You have a rather strange World view of success which is twisted by personal feelings.

    As the saying goes….facts don't care about your feelings.



  • Registered Users, Registered Users 2 Posts: 16,972 ✭✭✭✭osarusan


    It won't take it but it will fundamentally change it.

    Ai's ability to generate bland but reasonable academic texts will only improve and become less bland and more critical, and better able to locate, access, and integrate relevant citations and references. Student essay writing is changed forever.

    Already we are seeing institutions revert to class-based hand-written tests (which we moved away from for good reason), or coming up with other forms of assessment in which the primary aim is not to suitably determine whether a student has demonstrated understanding/mastery of content, but rather, that the assessment be 'AI proof'.



  • Registered Users, Registered Users 2, Paid Member Posts: 8,493 ✭✭✭plodder


    That's interesting. I'm curious about the size/scale of the subtasks that you give it. How much code would it typically generate for them? Also, can you say anything about the tasks themselves? How hard is it to describe them?

    I would agree that this approach is probably going to be more successful than asking for entire systems (or even most of an entire system) to be generated. Demos always make things look easy by using well chosen simpler "toy" examples that avoid all the complexities of real world systems.

    One of the reasons we're not allowed to use AI for our work product is I believe legal concerns about where the code is coming from. What if the code it generates is an exact copy of someone else's that they published on Github and was incorporated into the model? They might be a bit paranoid about that though.

    “The opposite of 'good' is 'good intentions'”



  • Registered Users, Registered Users 2 Posts: 43,749 ✭✭✭✭ancapailldorcha


    Most of those are dependent on lavish government handouts. He's best known as a pro-apartheid Nazi and you're shilling for him for reasons unknown.

    The foreigner residing among you must be treated as your native-born. Love them as yourself, for you were foreigners in Egypt. I am the LORD your God.

    Leviticus 19:34



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 54,592 CMod ✭✭✭✭magicbastarder


    if AI took my job and got me a redundancy package, bring it on…



  • Advertisement
  • Registered Users, Registered Users 2, Paid Member Posts: 8,493 ✭✭✭plodder


    I'd agree with all that. Yet, another thing I've noticed is how accurate Grok is on Twitter/X. For those still on it, it's worth looking at the answers it gives to people's questions. I haven't seen it give a wrong answer. Most surprisingly, only yesterday someone asked it to explain one of those trolling type posts that used irony to suck in people who took the words literally. Here it is.

    You can see a pile of responses along the lines of "You know, all those placenames are literally Spanish, DUH!!?" So, someone asked Grok to explain what she meant and it came out with this.

    Screenshot 2025-06-12 at 08.18.50.png

    There were other queries on the same thread before this one, that didn't even provide the hint that the post was a joke. It's hard to believe sometimes, that there isn't a team of 700 Indians behind it, it's that good.

    This is where I see AI having the greatest positive impact initially at least, on social media.

    “The opposite of 'good' is 'good intentions'”



  • Registered Users, Registered Users 2 Posts: 858 ✭✭✭BP_RS3813


    And who is paying for your life after those redundancy packages run themselves dry?

    AI is a cancer, should have been stopped 10 years ago with no one touching it.



  • Registered Users, Registered Users 2 Posts: 6,983 ✭✭✭eightieschewbaccy


    There's literally a Wikipedia dedicated to things Musk promised at certain deadlines that failed.. Then there's the failure of the cybertruck and the brands crashing sales. The reason I don't see him creating robots that can do complex tasks is because he's notorious for cost cutting. The reason full self drive with no driver involvement isn't coming any time soon is the absence of lidar for example.



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 54,592 CMod ✭✭✭✭magicbastarder


    my post was intended in jest and should not be overanalysed.

    not that i wouldn't welcome the redundancy money; it's not as if i'd put my feet up for the rest of my life.



  • Registered Users, Registered Users 2 Posts: 13,200 ✭✭✭✭bnt


    One thing I've noticed is the use of AI in recruiting. We're already seeing AI used for analysing CVs and doing screening interviews, so you can be turned down for a job without a single person in the loop. The person hiring for a role gets a shortlist only. Does that mean that someone looking for work should practice with AI, and get rated by AI, so they can get past the AI? Yup, a quick search found the likes of GreetAI offering such services.

    You are the type of what the age is searching for, and what it is afraid it has found. I am so glad that you have never done anything, never carved a statue, or painted a picture, or produced anything outside of yourself! Life has been your art. You have set yourself to music. Your days are your sonnets.

    ―Oscar Wilde predicting Social Media, in The Picture of Dorian Gray



  • Advertisement
  • Registered Users, Registered Users 2 Posts: 43,749 ✭✭✭✭ancapailldorcha


    I think we desperately need a Universal Basic Income honestly.

    The foreigner residing among you must be treated as your native-born. Love them as yourself, for you were foreigners in Egypt. I am the LORD your God.

    Leviticus 19:34



  • Registered Users, Registered Users 2 Posts: 858 ✭✭✭BP_RS3813


    Who is providing the income though? Some rich tech billionaire with his robots?

    If its the government, then where is the tax income coming from if half of us don't have jobs due to some AI or Robot?



  • Registered Users, Registered Users 2 Posts: 43,749 ✭✭✭✭ancapailldorcha


    I'd say a lot of money could be freed up by cutting off a certain South African's handouts.

    The foreigner residing among you must be treated as your native-born. Love them as yourself, for you were foreigners in Egypt. I am the LORD your God.

    Leviticus 19:34



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 54,592 CMod ✭✭✭✭magicbastarder


    i have heard about this; though in a comment on social media along the lines of 'recruiters are complaining that CVs are being written by AI now and all look the same, but since they use AI to read the CVs, fair game'.



  • Registered Users, Registered Users 2 Posts: 17,950 ✭✭✭✭Mr. CooL ICE


    I've described it badly by using the term "subtasks", but it would be creating local helper libraries. One example would be for parsing calls from an older, blackbox, system with with inconsistent conventions. Instead of getting booleans as true/false, it would sometimes be strings or 1/0. There's probably an open source library that could do this but I'd normally spend 10-15 minutes writing it to carer for every scenario I could think of, and then the unit tests on top of that. A one minute explanation to an AI tool with clear instructions of what I want and it gets created almost instantly, including unit tests and caters for edge cases I didn't think of (yes/no).

    More recently, I've found it great for getting pipelines up and running. We have a DevOps team that maintains the build tools and servers but our build pipelines are our responsibility. AmazonQ is great for generating cloudformation templates as long as you are explicit in your requirements but always needs tweaking to navigate our corporate infrastructure. And being someone with very little Linux experience, asking AI to create a bash script to do X, Y and Z is handy.

    I've also gotten AI to generate skeleton code for entire solutions by saying I want a DDD arch with mediator pattern etc, but I have never gone beyond an initial prototype; moreso to see if it could be done.

    The more I type, it feels like writing entire systems is like Tony Stark getting Jarvis to do the grunt work.



  • Registered Users, Registered Users 2 Posts: 371 ✭✭backwards_man


    The people here saying AI is sh1te or I tried it once and it didn't work exactly as I wanted, are assuming that AI is never going to get better. It will. It is. AI ability 5 years from now will be very different to its current capabilities. ChatGPT is not the cutting edge of where AI is now, it's just what AI companies are allowing the public to see and use.



  • Registered Users, Registered Users 2 Posts: 2,568 ✭✭✭skinny90


    I think AI will eventually replace work and jobs but not as rapidly as people say.

    I work in automation dealing with devs and testers mainly…

    A lot of breakthroughs and advancements will be how we can leverage AI for the better.

    For traditional companies those who embrace it, accept it and upskill will be the main beneficiaries.

    Those who resist will be replaced (eventually)

    I can see AI especially agentic creating a new burst of start ups given the productively outcomes, cost etc



  • Registered Users, Registered Users 2 Posts: 839 ✭✭✭poop emoji


    Be careful with cloudformation

    Several times I asked Amazon q to generate cloudformation config and it hallucinated config options in CF that simply don’t exist (yes I know the irony of Amazo product not knowing Amazon other products)

    The best part is that the config was very very believable and the particular param it invented would actually been great addition if it existed

    Other than that I noticed a marked improvement in last year, sometimes tests it generates are over verbose and at time won’t even compile the plausible sounding types it invents (despite being strongly typed language and it having access to the project) but yes I was also suprised at how it thinks of edgecases I wouldn’t have thought

    Tl:dr it’s getting better but still requires a squishy brain to check it hasn’t invented plausible sounding nonsense



  • Registered Users, Registered Users 2 Posts: 16,972 ✭✭✭✭osarusan


    Yes, it'll only get better, and as it understands more what people (in university at least) mean by 'critical thinking', it'll be able to produce work that satisfies that demand as well.

    I had a student wanting to do a PhD on Gen AI, and I advised her strongly against it, because whatever model and version she used to collect whatever data she'd collect - that version would be obsolete before her PhD would be finished, and whatever analysis she'd done about the capabilities and limitiations of Gen AI would be obsolete also.

    Post edited by osarusan on


  • Registered Users, Registered Users 2 Posts: 858 ✭✭✭BP_RS3813


    Whilst I agree that that he is a cnut, ejet and all of the above and would like to see that happen, will it ever actually happen though?



  • Advertisement
  • Registered Users, Registered Users 2, Paid Member Posts: 3,817 ✭✭✭Pa ElGrande


    I've heard that pitch from many AI salesmen. AI is the old computer maxim on steroids, garbage in, garbage out. When you throw data at AI searching for an answer ensure the answers are to be found within the dataset, otherwise it is going to spit out superficial looking garbage.

    AI output being sh1te is due to the technology being applied to situations that can't work or by people who don't understand what they are doing with it. Use cases where it can be reliably and economically used are not as wide as marketing people tell us. I know, I have to deal with said AI generated output on a daily basis where nearly everyone has bought into the hype and is afraid to say no. Luckily I have enough allies to push back on them, at the end of the day the results matter.

    AI is an investment bubble, the errors due to mis-allocation of capital will eventually be discovered, the bubble will burst, afterwards the technology will be applied to situations where it can be used profitably.

    To go back to the navvies versus bulldozer analogy again, you need a man to drive the machine, you need men to extract the materials and factories to build the machine, you need support services to maintain and use it in the field. These jobs did not exist before. AI technology needs people to maintain and manage it, as long as you can be productive with it, it will continue and develop. The energy requirements alone to support computer processing imply the application of the technology is grossly inefficient, that is a roadblock and limits it's viable use cases.

    Net Zero means we are paying for the destruction of our economy and society in pursuit of an unachievable and pointless policy.



  • Registered Users, Registered Users 2 Posts: 43,749 ✭✭✭✭ancapailldorcha


    No. Musk knows where too many bodies are buried.

    The foreigner residing among you must be treated as your native-born. Love them as yourself, for you were foreigners in Egypt. I am the LORD your God.

    Leviticus 19:34



  • Registered Users, Registered Users 2 Posts: 2,658 ✭✭✭Mr. teddywinkles


    Can I ask if all jobs are replaced by robots and ai, and no humans have an income how is an economy suppose to function. Do we scrap capitalism as we know it. Are these robots going to go to the shops to buy goods n services they produce. The whole thing is a fallacy.



  • Registered Users, Registered Users 2 Posts: 839 ✭✭✭poop emoji


    It’s ok I heard we will build 37GW of offshore wind and become Saudi Arabia of wind in Atlantic

    And all them datacenters that are being blocked from construction will randomly switch on and off depending on which way wind blows 🤣



  • Registered Users, Registered Users 2 Posts: 658 ✭✭✭FaganJr




  • Registered Users, Registered Users 2 Posts: 1,083 ✭✭✭techman1


    YYes Ai and even robotics will not be able to do those jobs. So humans will still be required to do the difficult and physically demanding roles. Demand for skilled trades men is likely to increase further.

    Covid caused alot of unforeseen consequences, by making many jobs so attractive by allowing working from home they made other jobs less attractive by comparison. Ironically it is those very working from home jobs that are now most threatened by AI



  • Registered Users, Registered Users 2 Posts: 371 ✭✭backwards_man


    AI is not going to replace all jobs. It's going to change the way some jobs are done.



  • Registered Users, Registered Users 2 Posts: 371 ✭✭backwards_man


    Not sure why you are replying to me on that. I agree with alot of what you said. AI is not going to replace most or all jobs. I never said it would. Just that it will be used a lot more to do functions that people do now. In certain roles what staff do in 10 yrs time won't be what staff do today. Roles will shift and change. But to those who say it's sh1te and doesn't work because the one time they tried it, they are in denial about how it will change workplaces.



  • Registered Users, Registered Users 2 Posts: 1,542 ✭✭✭floorpie


    AI is the old computer maxim on steroids, garbage in, garbage out. When you throw data at AI searching for an answer ensure the answers are to be found within the dataset

    You have the wrong mental model for LLMs, LLMs interpolate between data in training sets and can generate novel outputs. That's their strength. They aren't lookup tables or classifiers or similar.

    This is why they can already outperform humans on novel university exams, law, medicine, etc. I.e., the questions are not in the training set. Chain of thought and reasoning models are even more powerful for this. They can already generate novel and correct advanced research, codebases etc. And by 'novel', I mean that it doesn't exist anywhere in the world, nevermind in the training data.



  • Advertisement
  • Registered Users, Registered Users 2 Posts: 17,950 ✭✭✭✭Mr. CooL ICE


    On another note I don't think I've used stack overflow in months whereas it could have been used daily.



Advertisement