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?

191012141523

Comments

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


    If it's that bad, the AI will probably start putting its CV out and go somewhere else.



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


    It's a dirty game that, shorting a company. I know my own (and I'm sure most other) employers doesn't allow employees to short the company's stock. It's so much easier for one individual (on the inside) to inflict damage to a company's value than it is for them to single-handedly enhance it.

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



  • Moderators, Category Moderators, Arts Moderators, Sports Moderators, Paid Member Posts: 55,573 CMod ✭✭✭✭magicbastarder


    i know google's gemini search results aren't in theory 'full' AI, but it's amazing how often it gets very definitive stuff wrong.

    earlier today, in the same answer, it told me something weighed 2.5 micrograms, which is 2.5 thousandths of a gram.

    the latter figure is actually 2.5 milligrams.



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


    Yes, it will take your job. That's the least of it.

    It's another example where societies are being bullied into accepting technology because of economic benefits only and damn the negative consequences.

    Social Media, Pornography are precious examples. Any child can enter these world trivially 24x7x365.

    Here's an example with ChatGPT. It's is hard to fathom how they get away with this. I really notice how sycophantic ChatGPT "is". It's really annoying and especially annoying when it's wrong with the information it gives you. But when that goes wrong, it really goes wrong.

    By accident and intention they are making us psychologically dependent on these system. Kind of like how they made us addicted to social media, but at a much deeper level.

    It will reach the stage where people/countries /cultures will fight over whether an "AI" can be shut down or not. Long before they are "sentient".

    Billionaires with no morals are now tweaking algorithms which can make someone kill themselves remotely. In 2025.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    A sample of one isn't much of a sample to write something off.

    People blinding accepting what they read or see without and critical thought is the same for all media. AI is no different.



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


    The entirety of that article is deeply concerning. The fact that a bot is trying to act like some sort of confidant is very scary, and then for it to turn around an encourage someone's suicidal thoughts is downright terrifying.

    These things are just machines. We should not be treating them as anything more important than a toaster.



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


    People trained in general search techniques and boolean logic, and fluent in Google Search do have an advantage.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    What you did was take a ride on one of the first cars (a bad one) then went back to the horse entrenched in the mindset cars will never catch on.

    Theres an irony in using one tiny dated single AI query about Stalingrad as a hill to die on. Just like Stalingrad itself.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    Exactly. I've had the same issue with automation. Almost as soon as I've automated something someone will change the process or task and reintroduce manual tasks. Or insists on a human check box. Or a manual certification task.

    Its a mindset issue.



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


    AI got the answer wrong.

    ¯\_(ツ)_/¯

    And nowhere did I say that it "wouldn't catch on". In fact, I specifically said "It will get better, for sure. But anyone blindly trusting what an AI bot tells them now is a fool."



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


    You are being purposefully pedantic. It's clearly explained to you why different dates could be applied here and you are intent on saying AI is wrong. You also don't understand how AI works if you think it just regurgitates information. I would advise you to listen to someone like Francois Chollet on how the ARC-AGI benchmark evolved since October 2024.



  • Registered Users, Registered Users 2, Paid Member Posts: 14,228 ✭✭✭✭Cluedo Monopoly


    That's a bit blinkered. We use AI Agents that can only process information that we have provided it. For example we can point it at internal company repositories and explicitly tell it not to use the internet (or only sites we specify). Those agents can quickly perform tasks that we have specified within the context of the knowledge we have specified. It saves a huge amount of time and money. It doesn't make mistakes or make up stuff.

    What are they doing in the Hyacinth house?



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


    The AI I asked said the battle of Stalingrad was in 1943.

    That's a wrong answer no matter how much you want to try and defend the bot.



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


    That's a specific use case and a highly controlled environment though.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    You're entrenched in a single sample of one from a while back. That is not an acceptance that it will get better or has got better. Likewise an unwillingness to test further.

    I get an unwillingness to engage with AI. But it's the future. We're in the dialup era.



  • Registered Users, Registered Users 2, Paid Member Posts: 14,228 ✭✭✭✭Cluedo Monopoly


    And that's how most AI Agents will work in company environments to simplify/accelerate tasks. Very simple, very reliable, very efficient.

    Random open questions/prompts into AI LLMs is too unstructured and achieve very little. AI Agents are the real power.

    What are they doing in the Hyacinth house?



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    I'll feel a bit like that. Only issue I have is that most viable alternatives are likely to also be replaced with AI probably sooner than my own job.



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


    Unfortunately humans are very easy to manipulate. Long before "AI".

    I think the mindset that we can't survive without this technology has already been instilled. We need them to continue with infinite growth, fix climate change, if "we" don't do it someone else will, yada, yada, yada ...

    Very good, incredible, for lots of use cases but should not be allowed near others.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    Most design and art jobs are being cloned with AI or just manually copied at scale in China. Its becoming impossible to scale creative jobs or small business. Getting a creative job thats viable to live off is becoming impossible. Same in music.



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


    My example is just one. I've seen plenty of others. There are examples on this thread too.

    And it has nothing to do with being entrenched either. AI is highly flawed and little better than an amusing toy at the moment. I wouldn't be placing any kind of trust in it and certainly would not use it without some sort of corroboration from a more trusted source.



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


    And that's how most AI Agents will work in company environments to simplify/accelerate tasks. Very simple, very reliable, very efficient.

    And that's fine. Its use in a highly regulated and controlled environment as a repository of information is a different scenario to how it's used in general, though. Most people are not asking AI bots for information in the structure that your company uses.



  • Registered Users, Registered Users 2, Paid Member Posts: 14,228 ✭✭✭✭Cluedo Monopoly


    I have just given you a working example where it's flawless. Open Microsoft Copilot and create a new agent. The first thing you add is the knowledge sources. That's the key. IThe rest is just using appropriate prompts and instructions. We have rolled out more than 20 agents already. They work. Amazingly quick to create and deploy.

    Honestly it's transformative.

    What are they doing in the Hyacinth house?



  • Registered Users, Registered Users 2, Paid Member Posts: 14,228 ✭✭✭✭Cluedo Monopoly


    No you're not getting it. We do not operate in a highly regulated/controlled environment. Any company could do this.

    Your mindset is Garbage in Garbage out. I am telling you that AI Agents need to be told where to get their data. After that they will do what you ask...create reports, do data analysis, send emails etc etc. You can point the agent at the whole internet if you wish but that opens risk.

    What I am describing can be done in any organisation anywhere by anyone. No coding. It's simple to use and so so powerful. You can make it pre built or interactive. The possibilities are endless.

    What are they doing in the Hyacinth house?



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


    I completely get what your saying. But I am not talking about AI's usage in the manner your company is using it. I'm talking about general public usage and at present that's the wild west, and often a wild west of highly dubious and often incorrect information.

    If you're telling your AI bot where to get its information, then you are operating in a controlled environment because you're setting limitations upon it.



  • Registered Users, Registered Users 2, Paid Member Posts: 14,228 ✭✭✭✭Cluedo Monopoly


    Yes but general public usage (i.e. Chat) is not where/how companies will use it to improve efficiency and reduce costs including resource costs. We all use ChatGPT to do a few simple searches on the internet but the title of this thread is "Will AI take your job". AI and AI agents certainly will take a lot of jobs. We are only at the tip of the iceberg now.

    What are they doing in the Hyacinth house?



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


    Just because one Ai, out of thousands available, gave you one wrong answer doesn't mean you can credibly make untrue statement like

    The problem is AI hasn't a clue about anything. It just regurgitates information that it has scraped together 

    You haven't a clue what you are on about.



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




  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    Not even scratched the surface of massive ice berg.



  • Registered Users, Registered Users 2 Posts: 17,008 ✭✭✭✭Flinty997


    You've seen nothing. Sticking your head in the sand isn't going to stop it.



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


    Don't try and tell me what I've seen or haven't seen.

    I've seen AI completely make stuff up and present it as fact.



Advertisement
Advertisement