Noodleface
A Mod Real Quick
- 39,147
- 17,431
Sus. This is a huge topic for data centers right now. I think if they did use the lake there are environmental concerns with pumping heated water back into the lake.
IT/Software career thread: Invert binary trees for dollars.
- Thread starter Traak
- Start date
- 17,525
- 8,561
I was at a town meeting of sorts when I originally posted. Supposedly it's going to be 50k gallons of water and the heat exchange is just going to be food grade glycol. The system needs to be charged with water from...somewhere, but after that, requires little water.
Ironically, trying to decide which side to support feels like buying computer parts. I can struggle over every little detail like noise, light, electricity costs, environmental impact, tax benefits(both ways), job creation but I won't actually know until after the fact. Multiple people asked for various forms of promises or commitments on those things, but I wouldn't trust anything that wasn't legally binding. And even then, I don't trust the US's legal system and government wrt properly punishing bad acting corporations.
And I can't shake the feeling like this would be supporting disingenuous AI bullshit. Where's Copilot gonna install itself next because an AI datacenter was built in my town!?!?
Rangoth
Vyemm Raider
- 2,189
- 2,423
I’m working in a lab with some “developers” this week and got to see one of these. Code agents for the first time. I’ll admit it was neat and I could see how in good hands it would be helpful but holy fuck what I saw rustled my jimmies.
they literally cannot code. They keep asking the agent to do stuff and when it doesn’t it takes them an hour of asking the agent and struggling to know why when even I can see it’s using the wrong config or can’t parse the JSON. Whatever. Something simple and stupid. But the only thing they know how to do is keep asking the fucking agent. Of course, yes, H1Bs
they literally cannot code. They keep asking the agent to do stuff and when it doesn’t it takes them an hour of asking the agent and struggling to know why when even I can see it’s using the wrong config or can’t parse the JSON. Whatever. Something simple and stupid. But the only thing they know how to do is keep asking the fucking agent. Of course, yes, H1Bs
TJT
Mr. Poopybutthole
- 45,755
- 123,594
Interestingly I am part of the effort to understand AI costs and outputs for the company's AOP this year.
Finance is very very aware that we are on sweetheart deals with all of the AI companies right now but they are looking to project real $$ material gains from using AI vs its costs right now. While I think AI is extremely useful in various spots the increases in productivity certainly exist but many of them are not as exponential as you would think. We have a few production AI agents used across the company to do various tasks that have been identified as eating up significant time in the company. The agents have done well at just doing this for us as you go into ChatGPT and access the agent and it will retrieve things like specific invoicing, customer product usage, and other things and quickly summarize them.
While useful this ultimately ended up saving 50 man hours a month approximately. This isn't nothing, but it isn't replacing an entire person's salary either. When it comes to AI for coding it is 100% true that AI makes the coding part faster than it used to be. However if you look beyond that the majority of a senior engineer's time is not really spent heads down coding. It's researching and understanding before you actually execute it by writing the code to do whatever. AI's biggest and most advantageous gain according to our metrics is rapid POCing of unfamiliar technologies. Reducing the research time spent to narrow down the right avenue of development. As well as faster code completions in commits.
But not reducing overall feature development time by 50% or anything.
Finance is very very aware that we are on sweetheart deals with all of the AI companies right now but they are looking to project real $$ material gains from using AI vs its costs right now. While I think AI is extremely useful in various spots the increases in productivity certainly exist but many of them are not as exponential as you would think. We have a few production AI agents used across the company to do various tasks that have been identified as eating up significant time in the company. The agents have done well at just doing this for us as you go into ChatGPT and access the agent and it will retrieve things like specific invoicing, customer product usage, and other things and quickly summarize them.
While useful this ultimately ended up saving 50 man hours a month approximately. This isn't nothing, but it isn't replacing an entire person's salary either. When it comes to AI for coding it is 100% true that AI makes the coding part faster than it used to be. However if you look beyond that the majority of a senior engineer's time is not really spent heads down coding. It's researching and understanding before you actually execute it by writing the code to do whatever. AI's biggest and most advantageous gain according to our metrics is rapid POCing of unfamiliar technologies. Reducing the research time spent to narrow down the right avenue of development. As well as faster code completions in commits.
But not reducing overall feature development time by 50% or anything.
- 1
Noodleface
A Mod Real Quick
- 39,147
- 17,431
We had a VP come in and they were pumped about AI but what they said in the all hands was retarded
"AI is the worst it's ever going to be right now"
What they meant is AI is moving fast and everyday it gets better. But that saying is completely wrong. I'm sure she meant something else but I can't figure it out.
"AI is the worst it's ever going to be right now"
What they meant is AI is moving fast and everyday it gets better. But that saying is completely wrong. I'm sure she meant something else but I can't figure it out.