"You're reading it wrong..."

Please don't insult me. I can't read.

In all seriousness, for nearly all AI products, that's a good answer to the fuckwits that are using it inappropriately. Seriously, there are only a handful of people in the world I'd trust with a feature phone, let alone modern computers and

LLMs...

Not surprised. AI especially in the context is a series of answers looking for questions. And nobody knows what the questions are.

No shit.

However, what worries me is that AI is poisoning all of its 'deterministic/reliable/accurate' competitors, especially in areas like Internet Search.

Google and Bing are (deliberately??) Shit, to try to force people into using ChatGPT/Gemimi instead. Search indexing itself has become an impossible task due to an endless stream of machine-generated slop.

Jobs and techniques are being replaced by AI not because it does a good job, but it does a shit job fast, and you can re-roll the RNG until it gives you the bullshit answer you were looking for

AI in the form of LLMs like this are literally just statistical boxes deliberately biased to respond in a way that the creator considered favourable.

In essence, an LLM tells you what you want to hear, rather than the truth.

I have an entertaining half-hour every month of so when they release a new LLM that "fixes all the problems" where I try to get it to contradict itself.

It rarely takes more than a few minutes, and I can then have it running in circles between an obviously correct answer and the thing I'm telling it that "I want to hear".

Try it... it's quite revealing.

I usually try to pick something that it won't quite have a lot of details about in its training databases. One of them is to make up characters / actors in The Good Life (a 1970's sitcom, called Good Neighbors in the US). Within a few seconds, just by expressing frustration and denial with it, I can get it fabricating characters, actor's names, making up back stories, etc.

These things tell you what they've been trained you expect to hear from them. And there is zero intelligence in them.

It's a bit like a massive Google with an upvote button on it for each result. Within seconds you realise just what a terrible idea that is, for it to tell you what OTHER PEOPLE have told it you will want to hear.

I dunno, if someone kept asking me stupid questions, eventually I’d feed them some bullshit answer just to make them go away…

"an LLM tells you what you want to hear, rather than the truth"

Unlike the mainstream media. Faux News, the BBC, the Daily Heil, the Grauniad, Pravda, etc would never do that.

Not sure how you got upvoted, you sound like an idiot. First you're lumping in the propaganda arms of Rupert Murdoch, Lord Rothermere and the Kremlin with outfits that you (and I) may not always agree with on editorial stance, but at least value facts. But your implication is what - that we should get our news from some frothing ill-informed Nazi on youtube? Infowars? I realise that second one is redundant.

Is the AI bubble going to burst or just deflate like a shrivelled balloon?

DEEP THOUGHT:

I think the problem such as it was, was too broadly based. You never actually stated what the question was.

PHOUCHG:

B- b- but it was the Ultimate question, the question of Life, the Universe, and Everything.

DEEP THOUGHT:

Exactly. Now that you know that the answer to the Ultimate question of Life, the Universe, and Everything is forty-two, all you need to do now is find out what the Ultimate Question is.

My wife just upgraded her iPhone 13 to iOS 18. So the phone is too old to use AI apparently! But not too old to have Apple completely confuse her photographs. It took her ages to find a way to put them back into a simple date-ordered display. Bah, humbug!

If you were a bit cynical you might suspect that Apple's fucking of the Photos app in iOS 18 is just one step in a strategy to make their apps progressively crapper over the next year or so until Apple AI is the only way to get any use out of them, hence pushing people to run out and buy the newest iPhone just to be able to use apps like they did last year.

We used to speculate that Oracles products were built with the same design ethic.

Win the deal on features ticked on an RFP response with a UX so hideous that you'd then spend a fortune with their consultants to fix it.

Include your requirements about the UX into the RFP. See if they can meet that at an undercutting bid price.

It came down the Wifi in the latest update, and am about to turn it off. Annoying typos, annoying focus, totally annoying.

Clearly those behind it come from a Fail-Fast movement!

Fail-fast... and fix it, except there isn't a fix for AI apart from uninstall.

Most apps everywhere work pretty well if you only use them on superficial level. It's only when you start to delve into the lesser used features that they start to get crap. MS Word is a good example: it's great for the odd letter, flyers and party invites, but start using headers, page numbers, sections, outlining, indexing, etc. and things rapidly degenerate. These advanced features all work but it can become a frustrating wrestling match to make them work like they should have been intended to work. I assume it's because the devs never use the 'advanced' features; they just think they know how they are supposed to work and people who really need them buy pro software. Excel is probably the only exception to this rule that I can think of.

AI seems to start from the assumption that it both knows what you want to do and knows how to do it better than you. That might work for children and lazy management, but it feels like the start of a race to the bottom in terms of app utility as AI moves from just prompting you for what it thinks you want to to do to actively preventing you from doing things it doesn't understand. I'll be hanging on to my iphone 13 as long as possible, thanks.

Had this the other day. Its been a year or so since I used word to do anything particularly intense, I have a habitof doing anything like that on the Laptop where I can go sit somewhere nice. But last week I had to use it for some technical docs.

My Laptop has a locally installed copy of word and its old. My desktop, O365

Cue cries of WTF? Why? Will you stop bloody helping! Shortcuts that used to work, no longer do. Stuff that was simple (Formatting a sodding table) is now like pulling teeth and unintuitive.

I gave up, grabbed the laptop and did it that way.

I really wonder why MS are hell bent on breaking stuff for the sake of breaking it :(

I follow the maxim 'Just because something is AI-generated doesn't necessarily mean it is rubbish!'

A bit too subtle.

True, it just means it is extremely likely to be rubbish...

Or if you want a more comprehensive breakdown... it could have subtle, possibly undetectable errors to the layman. It could still read coherently and make sense but be completely wrong. It could expose an inherent bias in the LLM. It could be re-worded in a way that inadvertently changes the meaning or context.

Native speakers of a language (eg. English!) that have had decades of experience in its idiosyncrasies still encounter regular misunderstandings and misinterpretations. What chance has the infinite load of monkeys that is `AI`?

My Pixel 7 smuggled a "Smart Keyboard" feature in under the radar a little while back.

The first I knew of it was when WhatsApp kept suggesting responses for me each time I got a message from someone. But I couldn't even read the message that they'd sent me because it was covered up by the inane reply that my idiot phone thought would be the perfect response, which I had to close before I could even read what had been sent to me. Every fucking time. Seriously annoying.

It took a few weeks before I realised that it was actually my phone's OS sneakily inserting the responses into the App on my behalf - and not what I had assumed was a seriously poor UI choice by WhatsApp themselves.

I just got my Google Pixel 6a updated from Android 14 to 15 last night.

There was an awful lot of not very good "AI" products mentioned - which took a while to turn off. I've not particularly enjoyed my foray into Google's own 'Droid - there seem to be a lot fewer customisation options in Android than there used to be - and many more options for me to just trust Google to do all my settings for me. Which if I turn on seem to behave worse than I had the phone set up before.

I'm really looking forward to this bubble bursting. The hype is really annoying - but all the companies building piss-poorly thought out "AI" into everything makes setting stuff up extremely tedious.

Finally, what the fuck is wrong with these companies? I am getting heartily sick of being presented with an annoying pop-up for some piece of software they want me to enable, often right in the middle of me doing something important. How come your shitty AI didn't notice I was fucking busy you usefess fucking morons?!?!?! Only to be given the options install now or wait until later. Where's the fucking NO! option? And I definitely don't want your tour our new features option that takes me to a terribly animated video, with horrible background "music" slowly and painfully showing me all the "AI options we've added to delight you", that barely function.

Sadly Apple aren't much better. You used to have limited customisation options but at least a carefully thought through UI with minimal clutter and everything neatly organised. Now they're almost as much of an unfocused mess as Google or Microsoft.

The upside is that they are among the best supported among custom ROM producers.

Captain Hogwash,

I hadn't thought of that. The problem is I'm in the process of replacing this phone due to disappointing hardware. The USB socket has always been unreliable - cables appear to be able to move slightly from side to side, and even regular cleaning with compressed air, to keep the fluff out, doesn't seem to help much.

Also there's an intermittent error where the row of pixels in the middle of the screen suddenly all go clear. This sometimes lasts until I turn the screen off and on again, but once lasted for 4 days. Just as I was ordering a new phone, it stopped again - and didn't happen for a month. It's probably a loose connection - could be a bug, it's hard to tell.

Perhaps the answer is a newer Pixel and a ROM. Instead I've decided to give Nothing Phone a try.

Instantly turns your phone from a smartphone back into a dumbphone. Pretty much everybody could have predicted that this sort of shit was going to happen, and they were just asking for it with a name like that.

It's not about what's best for you. You are the product. This is about what's best for THEM. You are teaching their AI, and they are making gobs of cash off it rn, because you can't stop.

I'm not sure they are making cash off it. AI is a bottomless money pit right now (unless you make the right kind of hardware) and even in these fairly trivial manifestations, it doesn't seem to be living up to its billing.

Being that apple and google are in cahoots, my guess is everyone is being fed directly into the google hive mind. They won't do this unless it makes money.

They won't do this unless it makes money.

Google are totally fine with losing money. They fell, semi-accidentally, into a massive search and search-advertising monopoly that pretty much has funded Google ever since. They've then spaffed vast numbers of billions on side-projects in the hopes of making vast profits on that too - and pretty much failed.

In some ways they've been very far-sighted. They made a huge investment into Android, and yet make relatively small profits from it. But having their own phone OS has helped prop up their search monopoly (they were worried about local search being taken over by the phone OS makers - which would then give them an opportunity to move into general search. Also Android and mapping have gained them vast amounts of data - as well as helping maintain their dominance in search. Money well spent.

But they're spending untold billions on self-driving cars - something not even really linked to the industry they're in. That still looks like a black hole for investment with no payoff likely even within the medium term. I don't think Chrome OS has made them much cash either, or Google Suite. Though they make some money from Cloud Services - and they're in the datacentre industry anyway, so why not?

It also makes sense to go big on machine learning / "AI" - because they can't affford to miss out on it - if anyone can make it pay.

Self-driving cars on the other hand, seems not to be the most rational choice. A quick search suggests that Waymo has raised about $10bn in capital (most from Google/Alphabet) but some from VCs. Plus it's spent another $15 billion in running costs since 2016 - most of that also invested by Google/Alphabet but some of it will be receipts from customers - it doesn't make a profit, but does now have a significant number of daily customers. That's a hell of a lot of money - maybe $15-20 billion of shareholder's cash that they won't be seeing anytime soon. If ever.

Google are certainly willing to make big bets - with iminimal guarantees of future profitability.

Aspiring to achieve the intelligence of an apple.

70%+, across all platforms, don't trust pseudo-AI.

That goes a way to restore some of my faith in Humanity.

Cue the Hooray Henrys whinging about bias...

Can we leave Mrs Markle out of this please.

Read the article!

Auntie (i.e. the BBC to those who don;t know its nickname) did not use AI to generate the headline. Apple used AI to summarise an article published by Auntie and in doing so managed to completely screw it up in a way that only AI seems to be able to achieve.

after we slew the beast that was "big data"