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?

18911131423

Comments

  • Registered Users, Registered Users 2 Posts: 4,553 ✭✭✭arctictree


    Don't you still have to craft the questions to ask AI? And then you have to review, integrate and test the code it gives you.



  • Registered Users, Registered Users 2 Posts: 17,013 ✭✭✭✭Flinty997


    Kinda like we used to use xpertsexchange and then GitHub. I've been using it for code reviews and making suggestions on how I can do things better. Also it can comment my code which I then can tweak afterwards. I'm using it for a lot of the nice to do tasks I've never had time for. Or where mentoring is missing.

    Also useful to make sense of some of the specifications and requirements I get sent. A useful one is summarising the key points from a lengthy unstructured rambling meeting. Then creating a list of work items from it..



  • Registered Users, Registered Users 2 Posts: 17,013 ✭✭✭✭Flinty997


    I just asked AI and it replied...

    "...The Battle of Stalingrad lasted for about five months, from August 23, 1942, to February 2, 1943..."

    What I'm saying is how detailed and specific the question prompt really improves the answer.



  • Registered Users, Registered Users 2 Posts: 22,537 ✭✭✭✭Tony EH


    Asking the date of a battle is a pretty specific question with a pretty specific answer. There's nothing that vague about it. When I asked an AI the question "When was the battle is Stalingrad", it just simply got the answer wrong, because an AI doesn't know what it's saying. BTW, that was some time last year,so I would expect improvements to have been made in that time.

    However, the first contacts between the Germans and the Russians in the area of Stalingrad actually took place in July and by the 23rd of July Hitler had ordered the Army Group South to take the city. So even the response you got from AI isn't exactly accurate either.

    But the point still remains, in that you cannot trust the information that an AI gives you.



  • Registered Users, Registered Users 2 Posts: 17,013 ✭✭✭✭Flinty997


    A year ago is a lifetime in AI. It's like a flint lock to a fibre optic drone.

    The point is you can't talk to it like a human. It's not Star Trek. You need to learn how it works and how to write prompts. You either want to take that onboard or you don't.

    You don't just write one prompt either you refine it and work on developing the prompt.

    AI

    "...The first contact of the Battle of Stalingrad occurred on July 17, 1942, when German forces began their offensive to capture the city. The major offensive to seize the city itself began a few weeks later on August 23, 1942, but the initial skirmishes are dated to mid-July.

    July 17, 1942: The official start of the Battle of Stalingrad begins with the initial German offensive on the city's outskirts.

    August 23, 1942: The German 6th Army launched its major offensive to seize the city, leading to fierce urban warfare and aerial bombardment. ..."



  • Advertisement
  • Registered Users, Registered Users 2 Posts: 27,019 ✭✭✭✭Mrs OBumble


    Feel free to look up the pay rates for those jobs. Hell, you could even ask am AI tool to look them up for you.



  • Registered Users, Registered Users 2 Posts: 346 ✭✭purplefields


    It seems to be getting better and better at it. Now I do the most horrendous, sloppy requests, and it still seems to know what I'm on about. I do very cursory testing. less and less of it if I'm honest. I also use different AIs for different stuff as some appears to be better than others for specific things.

    I imagine AI testing is around the corner.



  • Registered Users, Registered Users 2 Posts: 14,752 ✭✭✭✭Frank Bullitt


    Back peddling I see. Did you get AI to make your argument for you, or do you just hold a snobby position over jobs that you look down on?



  • Registered Users, Registered Users 2 Posts: 10,146 ✭✭✭✭SuperBowserWorld


    Nail on the head.

    People love AI or don't care until it's about to take their job.



  • Registered Users, Registered Users 2 Posts: 927 ✭✭✭littlefeet


    Parts of the UK, Wales and Scotland, are peppered with professionally owned and managed Airbnbs or on other rental platforms as investments. For example, someone might buy a cottage in a scenic part of Wales or Scotland, a hiker's paradise, it's all managed from their computer, they use a cleaning turnover company who sends them back reports about the property, the property has the best of everything right down to the bean-to-cup coffee machine, its not a chancer hoping to make a bit of money, anything breaks they order a new one online and its diivered and the old one taken away, the cleaning company lets the delivery company in. Those staying often organise a food and alcohol delivery from a supermarket while staying. We stayed in two of them recently. The owner organised all this from London, Manchester, or wherever. That's modern life.

    30 years ago, how many Airbnb were there, how many cleaning companies were turning over Airbnbs, or even how many personal trainers were there or food delivery services where there?



  • Advertisement
  • Registered Users, Registered Users 2 Posts: 6,557 ✭✭✭silliussoddius


    Have you tried asking different LLMs the same question, or the same question to the same LLM at different times? It would be interesting to know how consistent they are.



  • Registered Users, Registered Users 2 Posts: 22,537 ✭✭✭✭Tony EH


    No. And to be honest I would have zero interest in doing so.

    I only asked that particular AI bot that question because I already knew the answer and was mildly interested in seeing what it would reply with. The point being if it could get that wrong and I knew the fact that it was wrong, I wouldn't be in any way confident about an AI answer about a topic that I knew little or nothing about.

    My issue with all of this is that there are too many people who just blindly accept what an AI throws out at them and I think that's a very dangerous road on numerous levels.



  • Registered Users, Registered Users 2 Posts: 553 ✭✭✭Anois_


    I'd say with 100% certainty that AI will take my job. My job is a full time mad bastard. Anyone remember seeing that on people on the dole's Facebook.



  • Registered Users, Registered Users 2 Posts: 553 ✭✭✭Anois_


    Anyway in all honestly unless they make AI robots my job is secure. It would actually be handy if there was robots as it would save me having to drill all the holes and hammer and screw in nuts.

    It would be years into the future before they get a AI robot to do my job. It would need to be able to move freely and with precision.



  • Registered Users, Registered Users 2 Posts: 929 ✭✭✭nolivesmatter


    "My issue with all of this is that there are too many people who just blindly accept what an AI throws out at them and I think that's a very dangerous road on numerous levels."

    Yeah that's the real problem there. It doesn't need to be perfect, it just needs to be good enough to get people to buy into it.

    I think people are wrong if they believe AI needs to be at least as good as a human to replace them. I think of it like video games. Back in the day when you bought one it was the finished product, but now they're pushed out incomplete and buggy as hell and maybe patched up at some point in the future. If companies can make the same or more money with less effort they'll do it , regardless of it reduces quality.



  • Registered Users, Registered Users 2, Paid Member Posts: 6,742 ✭✭✭TheIrishGrover


    My fear about AI is its foundation. These models have been built over years and many different iterations. People blindly accepting AI solutions to tasks without independent investigations. What's to stop a core flaw/vulnerability being accidentally or intentionally being in the very earliest days of certain outputs. I fear they could be susceptible to manipulation. One can already see how Twitter's AI is intentionally skewed in its answers.

    Let's just say you were a country who has a history of cyber warfare or hacking like Russia, China, North Korea or the US. AI is only as good as the info requested and supplied. You are such a state. Willing to play the long game. Willing to wait years. You "seed" info at massive scale (As a nation state could do) suggesting some encryption that you have breached. You make sure your target AI model recommends this as the gold standard in encryption. This gets adopted more and more and you have a snowball effect. AI model X suggests using this encryption. Soon enough it becomes the dominant encryption. Next thing you know, ALL AI models are saying "Well, 95% of fortune 500 companies use this encryption".

    Soon enough, you have a backdoor everywhere.



  • Registered Users, Registered Users 2 Posts: 358 ✭✭SpoonyMcSpoon


    I work in legal and can see already the benefits to using AI; will it take my job? No, but a lawyer fluent in AI would. However, as we know, the legal industry is quite good at protecting itself from dramatic reforms - the support staff are lokely to be the ones to lose their jobs unless, again, support staff trained in AI appear on the scene.



  • Registered Users, Registered Users 2, Paid Member Posts: 14,231 ✭✭✭✭Cluedo Monopoly


    The thing is that AI will be very easy to use. It already is and it's getting better all the time. You talk about fluency and 'trained in AI' but that's simply not required. AI is already easy to use, even the Agents. It's a bit like saying that people fluent or trained in Google Search will have an advantage.

    Post edited by Cluedo Monopoly on

    What are they doing in the Hyacinth house?



  • Registered Users, Registered Users 2 Posts: 346 ✭✭purplefields


    Why won't it take your job? I would have thought AI was the perfect match for legal stuff.

    You can put all legal texts that ever existed into a model. Retrieve any relevant case law on demand, properly interpreted.



  • Registered Users, Registered Users 2 Posts: 927 ✭✭✭littlefeet


    Because the likes of legal services will surround themselves with all sorts of legislation-based compliance stuff, which will make it mandatory for a human to do the work.



  • Advertisement
  • Registered Users, Registered Users 2, Paid Member Posts: 3,838 ✭✭✭Pa ElGrande


    image.png



    I'd be wary of building any process dependency on AI services until after the bubble has burst. There is the issue of what looks like wash trading between the major players to keep the bubble afloat. Eventually the likes of OpenAI and otherws will run out of other peoples money and fold. These companies are burning through cash by the pallet, the current prices they charge are loss making, once the correction happens everything gets repriced and if you have built a dependency on these services prepare to go under with them or pay whoever takes them over at a much higher price.

    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, Paid Member Posts: 3,092 ✭✭✭nachouser


    I work with a lad who makes handmade cards for birthdays, for him and his wife. He did it with co-pilot this time for a laugh. He took a picture of his wife standing with the background of a Castle in England they visited. He gave the name of his wife. Let's call her Mary. At one point, co-pilot created an image, unprompted, saying "Happy Birthday Jeffrey."



  • Registered Users, Registered Users 2 Posts: 7,686 ✭✭✭amacca


    After using free chat gpt on and off for the past year or so (varied tasks...lists of investment types, examples of 3x3 matrices, produce a download able file of a timeline in a specific style etc etc)

    I now think its being programmed to ask me clarifying questions right up to it's free use limit in the hope I'll buy a subscription......the pattern is there over the past couple of weeks....its like **** columbo...just one more question...right up to the point you hit the free limit..

    They either need to monetise yesterday or actually make a breakthrough that will generate the moolah...



  • Registered Users, Registered Users 2, Paid Member Posts: 8,827 ✭✭✭plodder


    It could certainly revolutionise that kind of legal research. But, it won't help so much with legal advice more generally or drafting contracts, or any other aspect of law that exists outside the public domain.

    An interesting question would be if it's able to apply existing law to new circumstances. Or what it actually comes out with if asked such a question …

    Post edited by plodder on

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



  • Registered Users, Registered Users 2 Posts: 3,900 ✭✭✭snotboogie


    When a battle "starts" can be an ambiguous question. Here is my interaction with GPT on your question:

    when was the battle of Stalingrad?

    The Battle of Stalingrad took place from 23 August 1942 to 2 February 1943.

    I mean from first contact between the two forces?

    Gotcha. If we’re talking specifically about the Battle of Stalingrad as a battle (not the whole Eastern Front), then:

    • First major contact in the Stalingrad area / approaches: late July 1942, as German forces of Army Group B pushed toward the city and clashed with Soviet defenders on the Don and in the outer defensive belts.
    • Heavy bombing and full-scale engagement focused on Stalingrad itself: from 23 August 1942, when the Luftwaffe massively bombed the city and ground forces reached its outskirts.
    • Intense street-to-street fighting inside the city proper: from around 13 September 1942, when German 6th Army entered central Stalingrad and the famous urban combat began.

    So, depending on how strictly you define “the battle” and “first contact,” you’re looking at late July 1942 on the approaches, or 23 August 1942 as the clear start of full contact in the Stalingrad battle itself.



  • Registered Users, Registered Users 2, Paid Member Posts: 3,200 ✭✭✭Hoop66


    The way things are going in my job at the moment, AI is fucking welcome to it.



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 55,574 CMod ✭✭✭✭magicbastarder


    amen to that.



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 55,574 CMod ✭✭✭✭magicbastarder


    has this been posted yet?

    https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-11-04-2025/card/michael-burry-is-back-with-two-big-shorts-palantir-and-nvidia-zQA6SOnar36KPg0BtGQS

    TL;DR - michael burry, the man behind the original 'the big short', has bet over a billion dollars on the collapse of palantir and nvidia share prices.



  • Registered Users, Registered Users 2 Posts: 844 ✭✭✭p15574


    I suppose that, unlike Fight Club, the first rule about taking a short position is you have to talk about taking a short position. Especially with his reputation. Try and encourage people to drop the stocks.



  • Advertisement
  • Registered Users, Registered Users 2 Posts: 22,537 ✭✭✭✭Tony EH


    Most historians agree that the battle of Stalingrad began on the 17th of July. But there's no real ambiguity involved with dates of most battles. The battle of the Somme was July 1st 1916, the battle of Jutland was 31st May 1916, the battle of Naseby was June 14th 1645, the battle of Agincourt was 25th of October 1415…etc. Stalingrad may confuse an AI because it was part of a larger operation, Fall Blau, but the AI didn't even mention that.

    Dating a battle has been a specificity for centuries, if not millennia. Trust me, I had to learn the dates of battles off by heart at one stage.

    So, to me, asking an AI about something as specific as the dates of something should be simple enough. The problem is AI hasn't a clue about anything. It just regurgitates information that it has scraped together and, quite often, that information is just flat out incorrect.

    It will get better, for sure. But anyone blindly trusting what an AI bot tells them now is a fool.



Advertisement
Advertisement