Artificial Intelligence

Started by seafoid, November 11, 2025, 07:15:55 PM

Previous topic - Next topic

Orior

Quote from: DaleCooper on November 11, 2025, 10:04:44 PMArtificial intelligence is impossible. Like "smart kettles" it's a marketing term that grew from scifi concepts.

It does a good job at mimicking intelligence and saves huge amounts of time.

In theory it should replace most of managerial flab and bureaucratic parasites.

That's exactly what an AI bot would say.
Cover me in chocolate and feed me to the lesbians

bennydorano

Quote from: Munchie on November 11, 2025, 10:33:24 PMWill AI tell those in the know that yearly pay rises without improving or accountability isn't the best idea?
I'd say even AI is familiar with the concept of inflation and cost of living.

Munchie

In the private sector you earn your pay rise, no such thing as a carte blanche pay rise for everyone no matter how productive or not!

brokencrossbar1

AI is very useful if you know how to use it. That is the key. Don't let it be the default,  have had several cases in work where claimants and clients have used AI to draft stuff and there have been laws and cases that don't even exist in their work. It is very good at assisting in drafting letters, defences etc but only as an aid not as the main option.

Always sense check, always double opinion. No matter what the old joiner phrase 'measure twice, cut once' will always apply.

lurganblue

Quote from: brokencrossbar1 on November 12, 2025, 11:12:41 AMAI is very useful if you know how to use it. That is the key. Don't let it be the default,  have had several cases in work where claimants and clients have used AI to draft stuff and there have been laws and cases that don't even exist in their work. It is very good at assisting in drafting letters, defences etc but only as an aid not as the main option.

Always sense check, always double opinion. No matter what the old joiner phrase 'measure twice, cut once' will always apply.

I've used it recently for some legislation research and it can be very hit and miss.  You definitely have to sense check it a lot and you can feel like you are going round in circles.  That said, it is very useful.

I find now that I use say chatgpt more than google when i want to find out something.  I like that you can go back in a conversation style to clarify, expand upon something or direct to something related.

imtommygunn

It's all about context. The mroe context it gets the better your outcome will be.

seafoid

It's good for work where there is a lot of context. The weakness is with hallucinations where there is little context. ChatGPT can't explain that there is no back information. Google translate has a similar problem translating fron English to Irish.

I wouldn't depend on AI for everything.

trueblue1234

It's mighty altogether. Sure JFK Jr is using it to write health reports in the US to justify his own opinion on health matters. Not just writing them, but also making up references when it didn't have any to back up these "findings".

Powerful stuff.
Grammar: the difference between knowing your shit

thewobbler

I reckon AI will bring us closer to the 1984 vision of the world than the Terminator one.

Algorithms can do anything data-related or process-related much more efficiently than a human being. Indeed they can even teach themselves how to be better at processing. But it's not possible for an algorithm's outputs to exceed the value of its inputs. 

And this our Catch 22. Who is it that is going to keep improving the inputs? And how can we trust that they're improving those inputs rather than controlling them?

We can't and we won't.


History will never have been more quickly rewritten by the winners.

Orior

Quote from: brokencrossbar1 on November 12, 2025, 11:12:41 AMAI is very useful if you know how to use it. That is the key. Don't let it be the default,  have had several cases in work where claimants and clients have used AI to draft stuff and there have been laws and cases that don't even exist in their work. It is very good at assisting in drafting letters, defences etc but only as an aid not as the main option.

Always sense check, always double opinion. No matter what the old joiner phrase 'measure twice, cut once' will always apply.

That's exactly what an AI bot would sa.....   oh never mind.
Cover me in chocolate and feed me to the lesbians

seafoid

https://www.ft.com/content/e4c9ac58-d64e-487a-b06d-e71be47f31c9

A GPT-4 model can use up to 463,269 megawatt-hours of electricity per year, according to research by academics at the University of Rhode Island, University of Tunis and Providence College. That is more than the annual energy consumption of more than 35,000 US homes. This demand reflects the expanding share of AI workloads in data centre electricity consumption. Global use of electricity by data centres is projected to more than double by 2030, and will reach about 1,800 terawatt-hours by 2040, enough to power 150mn US homes for a year, according to Rystad Energy.

smort


tbrick18

Quote from: smort on November 12, 2025, 03:10:36 PMDid you see the plan is to put data centres in space

https://www.theverge.com/news/813894/google-project-suncatcher-ai-datacenter-satellites

Brilliant idea.
Then when the aliens attack, they just have to scoop them up and we won't know how to do anything  ;D

In all seriousness, one of the biggest risks of AI I feel is that growing usage and reliance on it, will reduce the actual thinking carried out by real people. Rather than trying to work out how to solve a problem and learn as you go, just ask AI and trust blindly.

AI should only ever be an assistant in my view with a human in the loop for all content/decision based output. But, laziness and overconfidence will probably mean that human element could well become a thing of the past in some areas.

Banks of the Bann

Interesting discussion here on AI.

https://m.youtube.com/watch?v=b9EuLj5uPk0

Aleksandra Przegalinska, Senior Research Associate, Harvard Law School, and Vice-President of Kozminski University in Poland, warns that artificial intelligence is massively overhyped, is not replacing jobs or boosting productivity as promised, and the investment bubble could burst within a year.


Dag Dog

AI is being used as an excuse by firms to lay off people they over hired during Covid.
It makes them sound cutting edge by claiming they are using AI to power ahead with innovation. In reality, most of them are grappling with how to make use of it.