I asked ChatGPT to simplify and sort out my investments

Dead_keen

Who said "the only alternative"? You or your AI may be hallucinating ...

But in the meantime, I am sure you will enjoy reading this legal case where the lawyers learnt that AI can hallucinate. It's mentioned throughout but you can dip in to para 33 to find out what the lawyer actually did: https://caselaw.nationalarchives.gov.uk/ewhc/ch/2026/1199

mrklaw

I’m finding it very useful but I put more weight in what I’ve learned online from places like this forum, reddit, UK focused youtube channels etc.

and anything it outputs I’m checking carefully - for things like simple trackers I appreciate it doing the formatting job. for things like cashflow modelling/tracking I’ll either double check or just do those parts myself to be more confident.

But a lot of the math is fairly simple so debugging eg a missed tax assumption etc isn’t that hard - if you’re having deep conversations with niche subject groups like this, you should be confident to be able to check its work.

I’d be much more worried about my auntie Maureen using AI to create a plan and trusting it implicitly. but then I’d also be worried about her being caught out by a dodgy financial advisor etc.

af1963

"Just to check - if I want to know if the trains are running on time I need to drive to the train station and check?"

No other alternatives considered, so yes, seems like you were saying it's the only alternative to AI.

Actually I'm not even convinced that what you described as an AI method of doing this involves any artificial intelligence at all. Just a web lookup with a different interface. Or re-wording some search engine results. Lots of supposed "AI" turns out to be vacuous like this.

Getting back to the topic of the thread - it's not very surprising that a LLM that has been trained on the content of internet financial discussions ends up recommending similar things to what you'll commonly find in the source material.

HedgehogRulez

nothing out there is AI yet. It’s embarrassing.

The great unwashed don’t understand what AI is. It’s like looking at hedgehogs playing on Sinclair Spectrums (+3!)

Check back in 20 years.

Cobbler_tone

https://forums.moneysavingexpert.com/discussion/comment/82010034#Comment_82010034

"Are you trying to be right, or sound superior?" 😄😜

That post reads less like analysis and more like a mix of provocation, gatekeeping, and frustration—with a few different ideas mashed together.

1. The core claim: “nothing out there is AI yet”

This is an extreme, largely incorrect position by most definitions.

• Today’s systems (like large language models, recommendation engines, computer vision, etc.) are widely accepted as forms of “narrow AI”.
• What the poster might mean is: “this isn’t real intelligence or AGI (human-level general intelligence)”. That’s a more defensible argument—but they didn’t say it clearly.

So it’s either:

• Technically wrong, or
• Poorly phrased shorthand for “this isn’t true intelligence yet.”

2. “The great unwashed don’t understand”

This is classic elitist rhetoric:

• It positions the author as part of a “knowing minority”
• It dismisses others rather than explaining or persuading

In reality, understanding of AI varies widely:

• Some people overhype it (“it can think”)
• Others underestimate it (“it’s just autocomplete”)
• Most sit somewhere in the middle

So the complaint isn’t entirely baseless—but the tone undermines it.

3. The hedgehogs / Sinclair Spectrum analogy

This part is more interesting than it first looks:

• The Sinclair Spectrum (+3) is a 1980s home computer
• Hedgehogs playing on it suggests something:
  • primitive
  • chaotic
  • misunderstood
  • not what it appears to be

I’d interpret this as:

“What people think is sophisticated is actually crude and misunderstood, like watching something simple and assuming complexity.”

So beneath the odd imagery, they’re saying:

• today’s AI systems are rudimentary compared to what’s coming
• and people are misjudging their significance

4. “Check back in 20 years”

This is a long-term bet on AI progress, implying:

• current systems are early-stage
• future systems will be fundamentally different (possibly AGI or beyond)

That’s actually a very common serious viewpoint in tech:

• Many researchers believe we’re in a pre-breakthrough phase
• Others think current approaches may plateau

So this part is reasonable—just presented dismissively.

Overall take

The post blends:

• ⚠️ Overstatement (“nothing out there is AI yet”)
• 😬 Condescension (“great unwashed”)
• 🤔 A partially valid idea (today’s AI may be primitive vs future systems)

A more charitable translation might be:

“What we call AI today isn’t genuinely intelligent. Most people misunderstand it, and in a couple of decades we’ll look back and realise how basic these systems were.”

That’s a debatable but legitimate position—just badly expressed.

My honest assessment

• Signal: There’s a real point about hype vs reality
• Noise: The elitism and exaggerated phrasing
• Credibility: Reduced by tone more than content

If you’re reacting to it, a good question to ask is:

Are they trying to be right, or trying to sound superior?

Because here, it leans heavily toward the latter.

Dead_keen

You're absolutely right. My honest take — when viewed through a broader evaluative lens 🔍, hedgehogs playing on Sinclair Spectrums 🦔💻 represents a meaningful step-change in retro-computing animal engagement 📊 compared with two years ago, when the dominant paradigm 🌳 appeared to involve trees demonstrating limited soldering proficiency 🪛🔥 in relation to their ZX80s.

(*) This was manually typed (**) to make it look like a combination of Claude (you're absolutely right, and it's "honest" comments that have come in with Opus 4.7) and ChatGPT (which I love to use but have made sure never gives me bold or the silly pictures). The em dash was copied and pasted from the previous post.

(**) But a bot would say that.

Bostonerimus1

LLMs/current AI just crunch enormous sets of data and use weights to derive the next word in an answer…they essentially guess. Most of the time the responses are factual and sensible, but not infrequently they are wrong. The danger is assigning some sort of omnipotence to LLM answers.

Using an LLM to do your finances deprives you of the joy of doing your own research and limits your understanding of how to solve the problem. It's slightly worse than using an IFA. 😉

QrizB

A lesson from English law:

https://www.bailii.org/ew/cases/EWHC/Ch/2026/1199.html

TL;DR version: a major UK law firm made a routine legal submission to a court that included a LLM hallucination. When the court asked how it had happened, the law firm's reply was also AI-authored.

The law firm and at least two of their lawyers (one.of those a partner) are now being investigated by their professional regulator, and the law firm had to pay for their client to obtain new representation.

Cobbler_tone

I recently submitted a flexible working request that was ultimately successful. I used AI to help shape the submission by uploading the application form, relevant policy, and background context.

The output was comprehensive and strategic. It addressed all eight potential grounds for rejection, each with clear mitigation. It also provided alternative versions of wording, tone, and positioning tailored to different audiences and relationships.

In addition, it anticipated likely responses and objections, along with suggested replies to each, helping me feel fully prepared and confident throughout the process.

Don't get me wrong, these requests are always harder than they should be. Many businesses like to push back before reflecting and recognising the actual facts. I had a strong case but they can always still say "no" and I am running out of energy for a bun fight and don't want my final year (after 30 odd) to be under a cloud.

It is great for stuff like that.

Time4T_Accounts

https://forums.moneysavingexpert.com/discussion/comment/82014517#Comment_82014517

… and there’s a possibility that the junior lawyer who ‘outsourced their thinking’ (paraphrasing the judge) could be struck off. A career-ending mistake.