AI

[Polarbear]

This could be in either current affairs or geek.

AI

'It's going to be really bad': Fears over AI bubble bursting grow in Silicon Valley

https://www.bbc.co.uk/news/articles/cz69qy760weo

I'm confused on what they are saying about AI? Is it nothing like as useful as they thought? Is it just another dot com ponzi scheme? Why should we be worried? What does it matter if companies have over invested as that happens all the time in other industries.

Any answers in basic English please.
____________________
Triumph Trophy Launch Edition

[Easy-X]

Tech bubbles revolve around the next big thing to revolutionise the world, AI is just the latest. Money gets pumped into the gimmick, server farms get built and... then it doesn't sell as well as people expected. POP!

There's a couple of (so far) insurmountable problems with AI the main one being it lies and makes shit up. And we already outsource to India for that sort of thing The other, less obvious one is diminishing returns. I've seen a few mathematical analyses and from what I can tell these LLMs might approach human-level smarts but the amount of computing power to get there will need to be immense. The apocalyptic "and it became self aware at 2:14am" isn't gonna happen, not even close.

There will be jobs for the chop - a professor needs one less research assistant as the AI can write most of the paper - and I certainly don't envy graphic designers right now but at some point investors are going to see the lack of returns and move onto something else. POP!
____________________
Royal Enfield Continental GT 535, Husqvarna Vitpilen 401, Yamaha XSR700, Honda Rebel, Yamaha DT175, Suzuki SV650 (loan) Fazer 600, Keeway Superlight 125, 50cc turd scooter

[temeluchus]

[A100man]

[Ste]

When people talk about a bubble (like the dot-com bubble in the late 1990s), they mean that investment, hype, and expectations have grown faster than the actual technology’s ability to deliver results or profits.

So the worry isn’t that AI is fake or useless — it’s that too much money and optimism may be chasing too few real, sustainable business opportunities right now.

Companies like OpenAI, Anthropic, and Google are spending billions on data centers, chips, and energy. But most AI applications (chatbots, copilots, image tools, etc.) don’t yet bring in enough revenue to justify those costs.

Many businesses have tried using AI but find it hard to integrate or that it doesn’t improve productivity as much as promised. Some tools are great demos, but not game-changers in daily operations yet.

Startups that mention “AI” in their pitch are suddenly valued at absurd levels — even if they don’t have a working product. That feels reminiscent of the dot-com bubble, when everyone invested in anything with a “.com” in the name.

There aren’t enough GPUs, electricity, or engineers to sustain everyone’s growth plans. This makes costs skyrocket, which could hurt profitability if demand slows.

Even if AI itself continues to be transformative in the long run, a financial bubble bursting could cause:

- Massive layoffs in tech (as happened after the dot-com crash).
- Reduced funding for promising research or startups.
- A “winter” period where innovation slows because investors get burned.
- Knock-on effects in stock markets and the wider economy (since big tech stocks are heavily tied to AI optimism).

AI isn’t a scam or Ponzi scheme. The underlying tech is real and incredibly powerful — we’re already seeing major breakthroughs in productivity, science, and creativity.

The worry is more about timing and scale: people might be expecting too much, too soon.

In the long term, even after a bubble bursts, strong technologies usually survive (like the internet did after the dot-com crash).

[Easy-X]

For a real world example, my works have trained an AI to offer technical support 24/7 for our software. So that's a few Tier 1 employees we don't need for a start!
____________________
Royal Enfield Continental GT 535, Husqvarna Vitpilen 401, Yamaha XSR700, Honda Rebel, Yamaha DT175, Suzuki SV650 (loan) Fazer 600, Keeway Superlight 125, 50cc turd scooter

[Kawasaki Jimbo]

[Easy-X]

There's always something. Before AI everyone and his Chinese uncle were buying up graphics cards for Bitcoin mining and before that it was "The Cloud" (which has a least matured into something useful.)
____________________
Royal Enfield Continental GT 535, Husqvarna Vitpilen 401, Yamaha XSR700, Honda Rebel, Yamaha DT175, Suzuki SV650 (loan) Fazer 600, Keeway Superlight 125, 50cc turd scooter

[Polarbear]

[BusterGonads]

The thing about AI is that it is going to save a lot of money for businesses and it is going to make AI providers VERY rich indeed.

There are a lot of knowledge and communications jobs being done right now, in education, accountancy, medical fields, law and a lot more that can right now be done by AI very well indeed. These jobs will be going down the Swanny for sure. If people say that an implementation at their work didn't work very well, it just wasn't done properly. VAST numbers of jobs people do now will be done by machines in ten years time. History seems to suggest that new technologies see job losses, and then new job opportunities. Most jobs people do now did not exist a hundred years ago. Hopefully that trend will continue, but it is not certain. It just seems likely. The other thing is that when these kinds of revolutionary changes happened, they happened to the working class and manual workers. This change will happen to middle class and managerial people.
____________________
2023 Royal Enfield Classic 350; 2000 Honda CG125; 1997 Honda Nighthawk CB250

[stinkwheel]

It's being used to eliminate boring/repetative/unprofitable tasks from peoples schedules. What we should really be considering is if you can just leave an AI in charge of it and nobody notices much, do those tasks need to be done at all?

For example, marking my mandatory CPD and reflective learning log which I garauntee no human being has ever looked at. If they had, they would have contacted me to discuss its content. My reflective learning log has moved over the years from a tongue-in-cheek review of the quality of the lunch/snacks provided at each CPD event to a monologue citing several articles from peer-reviewed education journals as to why enforced reflective learning can be an unhelpful practice.
____________________
“Rule one: Always stick around for one more drink. That's when things happen. That's when you find out everything you want to know.”
I did the 2010 Round Britain Rally on my 350 Bullet. 89 landmarks, 3 months, 9,500 miles.

[Ste]

[hellkat]

Well, from an end-user viewpoint, I am finding Copilot quite useful, for both work and personal stuff.

It has cut a lot of tedium out of my day-to-day work, which is all that matters to me, and frees me up to do much more valuable projects.

Minute-taking is cut to a minimum, even though the meetings I have to attend haven't reduced. I'm still not directly saving the NHS bucketloads of money or saving lives but I feel like my NHS "footprint" has reduced - as a result of using Copilot I am definitely more productive.

But I did start to go down a Copilot rabbithole, from which I had to extricate myself as I started to feel like I was wasting time trying to understand stuff that didn't have much relevance. But its given me the opportunity to see that people are doing some interesting things to improve productivity - although whether its too little too late for the Gordian knot which is the NHS, I couldn't possibly say.

I'm sensibly invested in a tech tracker fund so I don't really put cash into specific tech companies, I'm too old to risk dotcom bubbles on technology which I'm not keeping up very well with any more (although I do try to keep up).
____________________
Not nearly as interesting in real life.

[panrider_uk]

[Easy-X]

Google keep throwing up their AI's interpretation of my search query and 9 times out of 10 it's bollocks I do use Grok a fair bit when I want an "executive summary" of things, just don't bother with anything political as it can only toe the party line.
____________________
Royal Enfield Continental GT 535, Husqvarna Vitpilen 401, Yamaha XSR700, Honda Rebel, Yamaha DT175, Suzuki SV650 (loan) Fazer 600, Keeway Superlight 125, 50cc turd scooter

[Hetzer]

I'm enjoying Ste's ironic-like use/abuse of AI in his posts.
____________________
"There's the horizon! Ride hard, ride fast and cut down all who stand in your way!"

[CrypticCrud]

I’ve never used it…

Or AFAIK I’ve never used it.
____________________
CBT - Aug24, Theory - Jun25, Mod 1 - Jul25, Mod 2 - Sep25
Kawasaki Z125 > Honda Z125

[struan80]

I had an Amiga 1200. It was amazing to me back then. Now look at us we're doomed. Will AI ever be able to feel emotion, really?

My use of AI so far is making funny photo's and listening to Alexa.

[Ste]

[Ste]

[struan80]

[temeluchus]

[Ayrton]

I think people were just too optimistic about it too soon and now it's going to a more reasonable level.

The whole thing can die for all I care. Every day some cunt in work says "I've asked chat GPT how to do this" and it's always the stupidest solution. Had someone tell me the other day he's using it to write emails for him so it gets the tone right. People cant even be bothered to do basic thinking anymore.

[Easy-X]

[panrider_uk]

I did use grok the other day to convert a written English query into quite a complicated (for me anyway) SQL SELECT statement
____________________
Current bikes: Honda ST1100 Pan European. Moto Guzzi V85 TT Travel