Data Centres

Pogue Mahone

Closet Gooner.
Joined
Feb 22, 2006
Messages
139,760
Location
"like a man in silk pyjamas shooting pigeons
Necessary evil or accelerants to the heat death of planet earth?

Just read this article about what’s happening in Ireland. Have to say AI is beginning to piss me off. The idea that we’re supposed to cut down on travel and use as little energy as possible while these cnuts churn through a whole city’s worth of energy so boomers can share AI generated memes or lobster Jeus on Facebook really boils my piss.

Or am I being unreasonable?
 
Pogue Mahone said:
Necessary evil or accelerants to the heat death of planet earth?

Just read this article about what’s happening in Ireland. Have to say AI is beginning to piss me off. The idea that we’re supposed to cut down on travel and use as little energy as possible while these cnuts churn through a whole city’s worth of energy so boomers can share AI generated memes or lobster Jeus on Facebook really boils my piss.

Or am I being unreasonable?
Click to expand...
It’s perfectly reasonable. Are we going to be better off in 50 years because of this? Possibly, but the likelihood is that we are going to be much worse off because of it.
 
Everything about AI is shite and will make everyone’s lives progressively worse.
 
Pogue Mahone said:
Necessary evil or accelerants to the heat death of planet earth?

Just read this article about what’s happening in Ireland. Have to say AI is beginning to piss me off. The idea that we’re supposed to cut down on travel and use as little energy as possible while these cnuts churn through a whole city’s worth of energy so boomers can share AI generated memes or lobster Jeus on Facebook really boils my piss.

Or am I being unreasonable?
Click to expand...

Most datacentres are still for cloud purposes rather than AI. So if most likely if you use the caf, or back up your photos you are complicit too...
 
Eugenius said:
Most datacentres are still for cloud purposes rather than AI. So if most likely if you use the caf, or back up your photos you are complicit too...
Click to expand...

Don’t get me started on cloud storage. Another example of “progress” that ended up being much more annoying than the technology it was supposed to replace. If it’s also setting the world on fire it can get to feck.
 
Rado_N said:
Everything about AI is shite and will make everyone’s lives progressively worse.
Click to expand...
Not everything. It makes my job astonishingly easier.
 
Rado_N said:
Everything about AI is shite and will make everyone’s lives progressively worse.
Click to expand...

100%. It’s strange how so many seem excited by the future, but nobody explains what’s going to happen people who lose their jobs to AI or companies make cuts as AI improves efficiency and processes. I don’t even think it’s that far away when this will start to happen, the bit that is frightening is that most type of jobs I can think of will be affected.

Recently I’ve been thinking a lot about social media and I’m sure there’s many positives from it, especially initially it seemed great. But now looking back over the last 12 years, I think I much preferred life pre social media- or maybe I’m getting old and miserable.
 
Pexbo said:
Not everything. It makes my job astonishingly easier.
Click to expand...
Yes, I've used it a fair bit when coding to simply get a skeleton of a script or method and then worked from that. It's especially useful when coding in an unfamiliar language.
 
Pexbo said:
Not everything. It makes my job astonishingly easier.
Click to expand...
In some areas like tax advisory it's pretty counter productive. If you put a pretty straightforward tax technical question into a supposedly advanced AI program it will give you a laughable answer. In fact you could say it is beneficial for some personal tax advisors who will get work dealing with the fallout from the crap that people have submitted to HMRC off the back of using AI instead of an agent.
 
Pexbo said:
Or we are more productive and take on more business?
Click to expand...

While employing the same amount of people. You see where this is going, right?

Anyhoo. AI decimating the job market is probably not for this thread. Which is mainly about the poxy data centres it needs to do its thing. I, for one, would much rather a world with more people employed writing code and less data centres doing work on their behalf.
 
Pogue Mahone said:
Don’t get me started on cloud storage. Another example of “progress” that ended up being much more annoying than the technology it was supposed to replace. If it’s also setting the world on fire it can get to feck.
Click to expand...

Cloud storage and 'the cloud' are very different.

I work in 'tech' and serverless architecture is just ubiquitous. Cheaper, far easier to maintain and you are absolutely fighting against pretty much every accepted standard to do anything else.
 
Pogue Mahone said:
Necessary evil or accelerants to the heat death of planet earth?

Just read this article about what’s happening in Ireland. Have to say AI is beginning to piss me off. The idea that we’re supposed to cut down on travel and use as little energy as possible while these cnuts churn through a whole city’s worth of energy so boomers can share AI generated memes or lobster Jeus on Facebook really boils my piss.

Or am I being unreasonable?
Click to expand...

A bit of whataboutism, but before cutting on data centers, that they have some utility, more in some cases less in others, we should be thinking to shut down cryptocurrencies that are 3-4 times more contaminant. 1% of worldwide emissions for useless speculative crypto is disgusting. At least data centers are useful in many areas
 
One thing i've never been able to wrap my mind around was the expansion of data centres for AI to generate art/cultural content. I just don't understand how people consume art and culture from a machine faking to be a human when literally the reason we buy a book, music or a painting is because it comes from another human that has created the work.

Then i saw @Sweet Square post on the Hypernormalisation thread. People are giving away the human connection to literally "have some noise in the background". I don't know if AI can really accomplish this with books as all your attention is in the book in front of you. Also, a lot of books are actually sold through connections to authors. But visual and sound art/culture is actually giving way to "filler noise" content and I don't know how to really feel about that. It's as if humanity could be on the verge of giving itself over to the the data centre/AI just so they don't have to be with themselves? And if that is the case, then there are genuine conversations to be had regarding the scope, intent and resourcing of AI/data centres. The inability or seemingly lack of desire to question why they are expanding, their purpose and demand for resources is an issue that will only increase with time.

According to a source close to the company, Spotify’s own internal research showed that many users were not coming to the platform to listen to specific artists or albums; they just needed something to serve as a soundtrack for their days, like a study playlist or maybe a dinner soundtrack. In the lean-back listening environment that streaming had helped champion, listeners often weren’t even aware of what song or artist they were hearing. As a result, the thinking seemed to be: Why pay full-price royalties if users were only half listening? It was likely from this reasoning that the Perfect Fit Content program was created.
Click to expand...

https://harpers.org/archive/2025/01/the-ghosts-in-the-machine-liz-pelly-spotify-musicians/
 
Last edited:
Pexbo said:
Not everything. It makes my job astonishingly easier.
Click to expand...
Copilot?

I love it but it’s quite worrying how quickly I became so reliant on it after 15 years in the industry. It’s definitely helped me problem solve tasks that would have taken a lot longer to scour the web for before but I use it now for such menial stuff I can’t help but think it’s having an affect there too.
 
Massive Spanner said:
Copilot?

I love it but it’s quite worrying how quickly I became so reliant on it after 15 years in the industry. It’s definitely helped me problem solve tasks that would have taken a lot longer to scour the web for before but I use it now for such menial stuff I can’t help but think it’s having an affect there too.
Click to expand...

Yeah it’s phenomenal to be honest. I’ve been using Edits for a while now. Just yesterday we had this DAL package in a monorepo that nobody wanted to touch and had been around for years, had next to no test coverage and little documentation. It was commonjs and made up of this huge interface that was being used all over the monorepo and then several other classes which were just wrappers for a CMS SDK. I renamed every file in the package to .ts then pulled them into the context and told it to refactor to a pure function approach. It took it about 40 seconds. Then I told it to add jsdoc headers to every function, another 40 seconds. Then I stripped out the old testing library and old tests and installed vite and vitest and told it to comprehensively test each file it had just refactored. This wasn’t perfect every time and took about 20 mins of adjusting and retrying as it had a bit of trouble mocking the CMS SDK but it got there in the end. I then traced where it had been used and refactored to the new approach. All in all about 30-40 minutes work and all existing unit and integration tests all passed. I reckon it would probably have taken about 4-5 days minimum if it was a manual developer task.

There’s now a 70 odd file pull request that’s going to be sitting there for years and years and nobody is going to want to touch.



Back on topic of this thread somewhat, my expectation is that cloud service providers will start using AI to evaluate the efficiency of the code their clients are deploying to their architectures and firstly start warning about inefficient code, eventually automating the optimisation process.

A lot of really shit code is written and deployed all over the place which is then executed literally millions and billions of times every second in data centres all over the world. It’s not really talked about outside of the developer side of the tech industry but it’s the equivalent of people living in uninsulated houses or having their heating on in the summer - it just burns energy.
 
AI is fantastic, just misused by most.

In clinical research and health care its already pushing boundaries nobody had dreamt of few years back. AI predicting protein folding is
so much faster than it was done before, which itself speeds up the creation of medications.

https://www.science.org/doi/10.1126/science.370.6521.1144

izzydiggler said:
Cloud storage and 'the cloud' are very different.

I work in 'tech' and serverless architecture is just ubiquitous. Cheaper, far easier to maintain and you are absolutely fighting against pretty much every accepted standard to do anything else.
Click to expand...

I don't necessarily agree. We have a hybrid environment, with a fully equipped and redundant datacenter on prem combined with Azure and I'd much prefer to keep our on prem servers because it's so much easier to maintain. IAC is great and all, but often tedious and the point that you are 100% dependant on a vendor that changes things around daily... the bigger the Microsoft ecosystem the buggier it's getting imo. Just a personal opinion of course.

At this point OpenAI, Microsoft, Google are already struggling with energy supply for AI that at some point users will have to pay for the currenly free chatgpt and data models to compensate for the energy cost. Let's see if they continue creating memes on it then.

AI could be called the biggest "invention" of our time, if used correctly. We need datacenters for it, we just need them to be greener.
 
We need more power to run the data centres? Well then let's build more power. There's solar, wind, nuclear, they all work. What's the problem. Just build stuff. I don't understand the mentality that says otherwise.
 
Pexbo said:
Yeah it’s phenomenal to be honest. I’ve been using Edits for a while now. Just yesterday we had this DAL package in a monorepo that nobody wanted to touch and had been around for years, had next to no test coverage and little documentation. It was commonjs and made up of this huge interface that was being used all over the monorepo and then several other classes which were just wrappers for a CMS SDK. I renamed every file in the package to .ts then pulled them into the context and told it to refactor to a pure function approach. It took it about 40 seconds. Then I told it to add jsdoc headers to every function, another 40 seconds. Then I stripped out the old testing library and old tests and installed vite and vitest and told it to comprehensively test each file it had just refactored. This wasn’t perfect every time and took about 20 mins of adjusting and retrying as it had a bit of trouble mocking the CMS SDK but it got there in the end. I then traced where it had been used and refactored to the new approach. All in all about 30-40 minutes work and all existing unit and integration tests all passed. I reckon it would probably have taken about 4-5 days minimum if it was a manual developer task.

There’s now a 70 odd file pull request that’s going to be sitting there for years and years and nobody is going to want to touch.



Back on topic of this thread somewhat, my expectation is that cloud service providers will start using AI to evaluate the efficiency of the code their clients are deploying to their architectures and firstly start warning about inefficient code, eventually automating the optimisation process.

A lot of really shit code is written and deployed all over the place which is then executed literally millions and billions of times every second in data centres all over the world. It’s not really talked about outside of the developer side of the tech industry but it’s the equivalent of people living in uninsulated houses or having their heating on in the summer - it just burns energy.
Click to expand...
Yeah it’s pretty incredible for writing tests, though I have found sometimes it can get the library completely wrong or write some absolutely pointless ones.

I also have concerns that if every developer in your company is just using an AI to write tests then is it actually passing quality checks properly? I’ve reviewed a lot of code in the last year that made no sense or had weird bits or didn’t follow our usual standards and that’s all cause of copilot, especially on junior devs.

Overall it’s obviously a massive benefit but I do think companies just handing it to hundreds of us unfettered will have long term problems.
 
nickm said:
We need more power to run the data centres? Well then let's build more power. There's solar, wind, nuclear, they all work. What's the problem. Just build stuff. I don't understand the mentality that says otherwise.
Click to expand...

Google “climate change”. We’re already struggling to meet existing energy needs with renewable energy. If that energy demand increases exponentially in a very short space of time then it’s obviously a huge problem for humanity.
 
WPMUFC said:
One thing i've never been able to wrap my mind around was the expansion of data centres for AI to generate art/cultural content. I just don't understand how people consume art and culture from a machine faking to be a human when literally the reason we buy a book, music or a painting is because it comes from another human that has created the work.

Then i saw @Sweet Square post on the Hypernormalisation thread. People are giving away the human connection to literally "have some noise in the background". I don't know if AI can really accomplish this with books as all your attention is in the book in front of you. Also, a lot of books are actually sold through connections to authors. But visual and sound art/culture is actually giving way to "filler noise" content and I don't know how to really feel about that. It's as if humanity could be on the verge of giving itself over to the the data centre/AI just so they don't have to be with themselves? And if that is the case, then there are genuine conversations to be had regarding the scope, intent and resourcing of AI/data centres. The inability or seemingly lack of desire to question why they are expanding, their purpose and demand for resources is an issue that will only increase with time.



https://harpers.org/archive/2025/01/the-ghosts-in-the-machine-liz-pelly-spotify-musicians/
Click to expand...

Great post. That’s what’s so depressing to me. This massive ramp up in energy consumption isn’t being used to make the world a better place. The opposite, if anything.
 
Massive Spanner said:
Yeah it’s pretty incredible for writing tests, though I have found sometimes it can get the library completely wrong or write some absolutely pointless ones.

I also have concerns that if every developer in your company is just using an AI to write tests then is it actually passing quality checks properly? I’ve reviewed a lot of code in the last year that made no sense or had weird bits or didn’t follow our usual standards and that’s all cause of copilot, especially on junior devs.

Overall it’s obviously a massive benefit but I do think companies just handing it to hundreds of us unfettered will have long term problems.
Click to expand...
100%. I think it probably works well out our company because we encourage a process where developers:

1. Develop as quickly as they can and get it to a working state.

2. Refactor, optimise and clean up.

3. Make sure it’s fully tested and documented.

4. PR review - these are always rigorous, at least two and one has to be by a developer as senior or more senior.

5. PR fixes.

6. Re-review and back to 5. if it doesn’t pass.

7. INT environment for QA testing

8. STAGE for UAT

9. PROD release.


So it fits in at 1,2 and 5 with plenty of layers of QA before it’s shipped. We also have a culture where developers only take responsibility up until it reaches 7, if something is shipped with an issue the blame lies with QA and the PR reviewers.

I think it works particularly well because a) it gives developers license to get stuck in and get the work done and also gives them the peace of mind that they aren’t going to be dragged into a meeting to explain their mistakes and b) it means that the code quality is generally very good because they know they won’t get it to INT otherwise let alone past it.

I can imagine that in agencies that focus on shipping products and forgetting AI is being massively abused. We have pretty comprehensive SLAs in place for all our products so having quality stable products means the amount of support we need to cover is kept to a minimum which means we can focus on new products. It’s a pretty solid model IMO.
 
It briefly mentions it in the article, but it isn;t just energy that these AI data centres consume. Its water too.

On the cooling front, Kate Crawford notes in an article published in Nature last week that a giant datacentre cluster serving OpenAI’s most advanced model, GPT-4, is based in the state of Iowa. “A lawsuit by local residents,” writes Crawford, “revealed that in July 2022, the month before OpenAI finished training the model, the cluster used about 6% of the district’s water. As Google and Microsoft prepared their Bard and Bing large language models, both had major spikes in water use – increases of 20% and 34%, respectively, in one year, according to the companies’ environmental reports.
Click to expand...

https://www.theguardian.com/comment...only-by-a-runaway-thirst-for-water-and-energy
 
Google wanted to build a data centre here in Chile, but luckily, some justice court stopped the project until they can prove its eco friendly.

The monster would need around 7.6 millions litres of fresh water per day. In a country that is slowly (quickly) becoming a desert (pretty much every other day I receive a record temperature alert on the weather app), It was always going to be a tough sale.
 
Habs said:
100%. It’s strange how so many seem excited by the future, but nobody explains what’s going to happen people who lose their jobs to AI or companies make cuts as AI improves efficiency and processes. I don’t even think it’s that far away when this will start to happen, the bit that is frightening is that most type of jobs I can think of will be affected.

Recently I’ve been thinking a lot about social media and I’m sure there’s many positives from it, especially initially it seemed great. But now looking back over the last 12 years, I think I much preferred life pre social media- or maybe I’m getting old and miserable.
Click to expand...
I’ve never had a social media account and never felt I was missing out on anything.
 
mirroring the argument in my original post, the lack of questioning by institutions and civil society is a big issue in the Data Centre / AI story. Interesting to see an uptick in these stories around the holiday time when less people are focused on news

OpenAI says it needs ‘more capital than we’d imagined’ as it lays out for-profit plan​

Click to expand...
https://www.cnbc.com/2024/12/27/openai-needs-more-capital-than-wed-imagined-moves-to-for-profit.html

OpenAI’s efforts to develop its next major model, GPT-5, are running behind schedule, with results that don’t yet justify the enormous costs, according to a new report in The Wall Street Journal.


This echoes an earlier report in The Information suggesting that OpenAI is looking for new strategies, as GPT-5 might not represent as big a leap forward as previous models. But the WSJ story includes additional details around the 18-month development of GPT-5, code-named Orion.
Click to expand...

https://techcrunch.com/2024/12/21/openais-gpt-5-reportedly-falling-short-of-expectations/

AI data centers are multiplying across the US and sucking up huge amounts of power. New evidence shows they may also be distorting the normal flow of electricity for millions of Americans. This map shows readings from about 770,000 home sensors, with red zones indicating areas with the most distorted power An exclusive Bloomberg analysis shows that more than three-quarters of highly-distorted power readings across the country are within 50 miles of significant data center activity. While many facilities are popping up near major US cities and adding stress to already fragile grids, this trend holds true in rural areas as well.
Click to expand...

https://www.bloomberg.com/graphics/2024-ai-power-home-appliances/
 
I believe the growth/impact will have diminishing return while the cost will increase exponentially. Just like crypto mining. And I hope at that point all investors will feck off and stop growing AI.
 
Habs said:
100%. It’s strange how so many seem excited by the future, but nobody explains what’s going to happen people who lose their jobs to AI or companies make cuts as AI improves efficiency and processes. I don’t even think it’s that far away when this will start to happen, the bit that is frightening is that most type of jobs I can think of will be affected.
Click to expand...
We are already losing jobs to it. My company now has stopped hiring casuals to do mostly admin stuff that is now automated or made so easy as to not require extra staff. And this is just the start.

There is nothing much that can be done to stop it, just as with machinery/automation reduced employment in weaving/cloth making and agriculture during the Industrial Revolution. This is why we should have started exploring Universal Income a long time ago.
 
Last edited:
You must log in or register to reply here.