We’d like to remind Forumites to please avoid political debate on the Forum.

This is to keep it a safe and useful space for MoneySaving discussions. Threads that are – or become – political in nature may be removed in line with the Forum’s rules. Thank you for your understanding.

📨 Have you signed up to the Forum's new Email Digest yet? Get a selection of trending threads sent straight to your inbox daily, weekly or monthly!

Doom scrolling and the next financial crash!

1…1011131516…19

Comments

  • aroominyork
    aroominyork Posts: 4,170
    Part of the Furniture 1,000 Posts Name Dropper
    Forumite

    Methinks you guys should take this to a tech forum. Now, about those Japanese smaller company funds...

  • Bostonerimus1
    Bostonerimus1 Posts: 2,272
    1,000 Posts Third Anniversary Name Dropper
    Forumite

    I've asked Gemini cross border taxation questions and the answers are often vague or just wrong. I put this down to the AI being taught on all the bad answers posted to Reddit which drown out the more accurate answers of professionals and a few knowledgeable amateurs. Reddit content should get a very low weighting in the fitting.

    And so we beat on, boats against the current, borne back ceaselessly into the past.
  • GeoffTF
    GeoffTF Posts: 2,892
    1,000 Posts Fourth Anniversary Photogenic Name Dropper
    Forumite

    You can misuse AI, but that does not mean it is useless. It is improving rapidly.

  • Bobziz
    Bobziz Posts: 753
    Sixth Anniversary 500 Posts Name Dropper
    Forumite

    Absolutely, and like any tool you need to learn how to use it. For tax questions I tend to ask it to use only HMRC sources.

  • Bostonerimus1
    Bostonerimus1 Posts: 2,272
    1,000 Posts Third Anniversary Name Dropper
    Forumite
    edited 5 September at 2:38PM

    The problem is AI comes with an aura of perfection and many people will take its answers as correct 100% of the time.

    The LLMs themselves are showing marginal gains for large training costs, but the "harnesses" to apply AI are getting more useful. I assume AI will get embedded into many things and there will be ups and downs. I work with a software engineer that has used Claude Code to port some routines from one DAQ board to another and it worked very well. However, the actual code isn't easy to follow and he's not sure how efficient it is and is worried about maintaining it.

    I'm not completely against AI, but I do find the current crop of AI CEOs to be scary and the way it's being financed and the scale of the infrastructure being built seems unsustainable to me.

    And so we beat on, boats against the current, borne back ceaselessly into the past.
  • Bostonerimus1
    Bostonerimus1 Posts: 2,272
    1,000 Posts Third Anniversary Name Dropper
    Forumite
    edited 5 September at 2:40PM

    I've used Gemini for US/UK cross border tax questions and most of the time it's right, but it often gets some details wrong. A while back it had not caught up with a new HMRC policy to tax lump sum distributions from US DC retirement account. It eventually started to get it right, presumably after new training. However, I would use a cross border tax specialist AI that is deeply trained on HMRC, IRS and professional tax specialists data and was guaranteed to be up to date.

    I don't think the future is General Intelligence, but it might be Specialized Intelligence.

    And so we beat on, boats against the current, borne back ceaselessly into the past.
  • Dead_keen
    Dead_keen Posts: 498
    Part of the Furniture 100 Posts Name Dropper Combo Breaker
    Forumite

    For the oldies among us, as translated by ChatGPT:

    1. An MP3 player is a bit like a record player.
    2. Like an album cover, songs are displayed in the order that they are laid down.
    3. @talexuser would prefer to have the songs in alphabetical order.
    4. There is a bit of a disagreement about how to change the order of the songs.
    5. In the old days, you could record the songs onto reel-to-reel tape, cut the tape between songs with a razor blade and splice the pieces back together in a different order. Voilà: a new running order.
    6. The disagreement is essentially about whether AI suggested the wrong digital equivalent - or suggested the right one but the user ran it incorrectly.

    If it was me, I would open Claude Code or Codex and say: Find the songs on my computer and do whatever you need to do to so that my MP3 player plays them in alphabetical order. (***)

    (*** an over-simplification - I'd ask it to find me a better MP3 player that allows different ways of sorting songs (as the Phantom of The Opera doesn't work for me in alphabetical order) but, failing that, I'd (i) tell it where the files were as I don't want it rummaging all over the place, (ii) test it on one album first, and (iii) not delete anything until I am happy it works properly.

  • masonic
    masonic Posts: 30,830
    Part of the Furniture 10,000 Posts Photogenic Name Dropper
    Forumite
    edited 6 September at 2:18PM

    I'd like to apologise for taking the bait. I shall delete my off-topic posts. It is now clear to me that there is an intent to derail this thread and unwillingness to engage in the proper place. Nevertheless, I have created the following thread for the OP to continue discussing with an audience that is more likely to be helpful.

  • masonic
    masonic Posts: 30,830
    Part of the Furniture 10,000 Posts Photogenic Name Dropper
    Forumite

    Quoting this over here on Savings & Investments because it is about the AI bubble and whether your experience is evidence that AI is over-hyped, and the valuations associated with it are unrealistic.

    Using a technical edge-case as evidence that the AI sector is in a bubble does not make any sense. The commercial value driving valuations right now comes from automating large-scale workflows, processing huge amounts of data, writing standard code, and streamlining operations across industries. Pointing to a consumer AI search failing to solve a hyper-specific, obscure utility task doesn't prove the technology lacks economic value, any more than a Swiss Army knife failing to act as a sledgehammer proves multi-tools are useless.

    What makes the argument even weaker is that conventional search engines, technical forums, and existing software utilities haven't handed over a quick, off-the-shelf solution to your problem either. When a problem is genuinely niche or fundamentally constrained by how a system works, neither a standard Google search, an AI summary, nor human forum members are going to deliver an easy answer. You have your thread in the techie stuff forum now, so maybe someone will have some expert knowledge and make a suggestion.

    I am actually in the camp that believes AI company valuations are currently stretched, overly speculative, and due for a market correction. However, basing a macro-economic bear case on "it couldn't solve my specific tech issue" does the position a real disservice. There are plenty of solid financial, ROI, and capital expenditure arguments for why AI stock prices may be inflated, but this just isn't one of them.

★ ★ ★ Meet your Ambassadors

🚀 Getting Started

Hi new member!

Our Getting Started Guide will help you get the most out of the Forum

Categories

  • All Categories
  • 355.8K Banking & Borrowing
  • 254.9K Reduce Debt & Boost Income
  • 456.2K Spending & Discounts
  • 248.3K Work, Benefits & Business
  • 605.9K Mortgages, Homes & Bills
  • 179K Life & Family
  • 263.8K Travel & Transport
  • 1.5M Hobbies & Leisure
  • 16.2K Discuss & Feedback
  • 37.7K Read-Only Boards

Is this how you want to be seen?

We see you are using a default avatar. It takes only a few seconds to pick a picture.