• 0 Posts
  • 425 Comments
Joined 1 year ago
cake
Cake day: June 22nd, 2023

help-circle






  • It’s wild that Amazon wants RTO so bad when they’re famous for a philosophy of not letting teams talk to each other directly, and instead forcing every team to publish APIs for their services that other teams can consume.

    I mean, sure you could still be collaborating internally with your team but they seem ideally suited to take advantage of remote work.

    My guess is that for a lot of big tech companies, half the RTO push was just because they didn’t want to have to look like they overspent so much on super fancy lavish offices now that everyone was suddenly happier at home.



  • Microsoft posted a revenue of $211.9 billion for 2023. Keeping in mind that the vast majority of the world’s population uses Microsoft Windows and Microsoft Office and loads of online applications run on Microsoft Azure, the economic impact of Microsoft’s products is probably counted in trillions of dollars.

    Comparing this to countries with the same ballpark energy consumption, Azerbaijan’s GDP was about $78 billion, Slovakia’s GDP was around $127 billion, and Iceland’s GDP was approximately $30 billion in 2023.

    The economic output of Google and Microsoft by far exceeds these countries’ GDPs, highlighting the vast financial scale of these tech giants relative to their substantial electricity consumption.

    Oh yeah, it’s crazy to think that! I don’t know where I would have gotten that idea, other than the article that OP linked that we’re all discussing.

    Yes, training new AI models uses a bunch of power, so does building out any new infrastructure. Atleast Microsoft and Google use a far high percentage of renewable power than most other industries.




  • As of last year ~70% of software developers were using copilot or a similar AI assistant. The legal field has seen a drop off in junior hires because of AI assistants. Snapchat’s AI filters and tools have long been a huge draw for that platform (and then copied by everyone else to avoid bleeding users), and Bing saw massive user growth after integrating OpenAI.

    AI has problems and limitations but it’s absurd to think there’s no demand for it just because it’s pushed by annoying people. Everything with hype will get pushed by annoying people.



  • You also use Gmail and force Google to run their servers to power it.

    Reducing your carbon footprint as much as possible is important, but it’s absurd to get mad at companies that power 90% of the world’s businesses for using a bunch of power to do so. It takes power to do those things. Get mad at the companies who are over consuming relative to their peers and those that are driving demand towards unattainable activities. Just getting mad at people for moving and using energy is absurd.



  • No, it’s not.

    Them making money implies that they are being paid to use power, which is true. Their absolute carbon footprint is irrelevant given that most of what the carbon they use is at the request of someone else. The metric to judge them on is their carbon footprint relevant to peers.

    I.e. it’s not fair to judge a cab company for driving someone somewhere (judge the person choosing to hire a cab), but it is fair to judge them if they use gas guzzlers instead of EVs.


  • It makes more sense than trying to build Master Chiefs. Robot powered body armour sounds awesome, but in reality it’s a lot more efficient to avoid getting hit then to try and take hits and keep moving, especially if you’re a rich country with an advantage in micropocessing resources (assuming Taiwan doesn’t fall).

    What they’re talking about isn’t crazy either, build out a platform that can handle robust military comms, connect it to a modern military information network, and then have local machine learning capabilities to handle local on site applications.

    I mean, even just a bunch of soldiers in a line with networked helmets would create a line array of microphones that could theoretically be used to figure out where enemy fire is coming from / how many people are shooting at them, what type of gun is firing at them, how much they have left, etc. If humans today can pick out where a sniper might be shooting them from, Im willing to bet a computer connected to an array of different soldiers’ sensors will be able to do it not too long for now.