Felt like Sasha was dodging questions about the environmental effects of climate change. The email didn’t ask about personal culpability- they wanted to know the environmental effects. And skipping the question about Google search vs AI church was also frustrating. It just felt like she was trying to subtly defend AI through, idk, diffusion of responsibility and a vague attitude about the environmental effects I think Kevin just made a good point about context though 45:57 - Then Sasha answered in a frustrating way
I think her response was reasonable in that we live in a world in which every day corporations waste more energy, water, etc, than an individual could in their entire lifetime even if they were trying. There’s no utility in feeling guilty about your behaviors when changing them would not make a difference. Even if everyone tried their hardest to stop using AI, those data centers and compute resources would not go away, they would just be used in some other way. Believing you can make a difference as an individual can be empowering, but most people seem to use it as a way to feel guilt and shame, which isn’t healthy, helpful, or necessary.
26:00 So, the President (Trump, specifically) is now allowed to just decide on a whim against the decisions of Congress AND the Supreme Court…? Is that where we’re at now, with this possibility?
Zuck larping as a manly man was not on my 2025 bingo, but here we are. The only way this turns into a positive is if it motivates him to have a cage fight with Musk
The enviromental comparisons are probably old. GPT-4o Mini is over 60% cheaper than GPT-3.5 Turbo for token processing and the models will only get more energy efficient over time until they run locally and then they will consume even less energy
Agreed. I was aware that training LLMs required a lot of energy; I wasn't aware of the energy impact of my own stupid ChapGPT/Gemini queries. I am chastened😥. I strive to be more judicious in the future.
I would actually pay good money to see the look on their faces, when the Red Note refugees realize they got banned from the app for posting pictures of Winnie-the-Pooh
Why are you repeatedly comparing the environmental costs of searching for info on an LLM to searching on Google? The vast majority of Google searches now start with an LLM query. So literally almost every time you "Google" something, you are also prompting an LLM, right?
Well we should be very concerned about the environmental impact of AI, the companies building the solutions are driving efforts to create sustainable and less expensive energy sources. Also, these tools are going to be very helpful in improving a range of environmental problems and understanding which environmental factors are really impactful and which are a by-product of assumptions.
HOW are these tools going to be more helpful? We know what environmental factors are most impactful-it’s fossil fuels. This is not a problem that requires advanced computing, it requires a step decline in the amount of carbon we are putting into the atmosphere, and we have a limited amount of time to do it. Please explain to me how doing the opposite of that is going to help the problem?
Regarding AI and energy use, the companies will never release accurate statistics around energy use in the current political climate. Urban legends will be more effective even if they are not precise. Of course, researchers should still pursue accuracy, but if current trends hold research will make little impact on policy or public perception in the coming years.
To female employees at Meta - your thoughts on Zuck's comments? Or is the money coming in so good that you (just like white women this past election) look the other way and justify/excuse his words?
I loved this podcast until today. When people start to misrepresent dialogs of people they don't like on purpose and clearly show an excerpt of the dialog that has nothing to do with what they are actually inferring, that's when any person with principles has to move away. Thank you for your tech related content until now but I'm unsubscribing today.
Felt like Sasha was dodging questions about the environmental effects of climate change. The email didn’t ask about personal culpability- they wanted to know the environmental effects.
And skipping the question about Google search vs AI church was also frustrating.
It just felt like she was trying to subtly defend AI through, idk, diffusion of responsibility and a vague attitude about the environmental effects
I think Kevin just made a good point about context though 45:57 - Then Sasha answered in a frustrating way
I think her response was reasonable in that we live in a world in which every day corporations waste more energy, water, etc, than an individual could in their entire lifetime even if they were trying. There’s no utility in feeling guilty about your behaviors when changing them would not make a difference. Even if everyone tried their hardest to stop using AI, those data centers and compute resources would not go away, they would just be used in some other way. Believing you can make a difference as an individual can be empowering, but most people seem to use it as a way to feel guilt and shame, which isn’t healthy, helpful, or necessary.
26:00 So, the President (Trump, specifically) is now allowed to just decide on a whim against the decisions of Congress AND the Supreme Court…? Is that where we’re at now, with this possibility?
I hope not
Behind the scenes footage at the end was hilarious
Here to see if there’s video material for the red note quotes that made me crack up listening to the podcast. Great pick Kevin
Zuck larping as a manly man was not on my 2025 bingo, but here we are. The only way this turns into a positive is if it motivates him to have a cage fight with Musk
Kevin is always giving a daddy vibe. But Casey you are aging like a fine wine
Thanks guys, great video!
The take home message from the guys is to listen to Mark Zuckerberg and "Man Up!" Too funny! 😂
So, so good. All the accountants in the MMA department must be named Marshall.🤣
Woefully uninformed question: Is there any reason why data centers can't use grey water?
The enviromental comparisons are probably old. GPT-4o Mini is over 60% cheaper than GPT-3.5 Turbo for token processing and the models will only get more energy efficient over time until they run locally and then they will consume even less energy
Agreed - running 7 to 70B models locally with Ollama and my MacBook still hasn't asked for any water. :)
Brilliant information about the environmental effect
Agreed. I was aware that training LLMs required a lot of energy; I wasn't aware of the energy impact of my own stupid ChapGPT/Gemini queries. I am chastened😥. I strive to be more judicious in the future.
Great podcast guys and great insights, thank you!
Love you guys!!!! ❤
I really enjoyed this episode! 💕🌈
Not seeing enough gold chains
Moar chains
I would actually pay good money to see the look on their faces, when the Red Note refugees realize they got banned from the app for posting pictures of Winnie-the-Pooh
Does congress even matter anymore?
Why are you repeatedly comparing the environmental costs of searching for info on an LLM to searching on Google? The vast majority of Google searches now start with an LLM query. So literally almost every time you "Google" something, you are also prompting an LLM, right?
Well we should be very concerned about the environmental impact of AI, the companies building the solutions are driving efforts to create sustainable and less expensive energy sources. Also, these tools are going to be very helpful in improving a range of environmental problems and understanding which environmental factors are really impactful and which are a by-product of assumptions.
HOW are these tools going to be more helpful? We know what environmental factors are most impactful-it’s fossil fuels. This is not a problem that requires advanced computing, it requires a step decline in the amount of carbon we are putting into the atmosphere, and we have a limited amount of time to do it. Please explain to me how doing the opposite of that is going to help the problem?
Regarding AI and energy use, the companies will never release accurate statistics around energy use in the current political climate. Urban legends will be more effective even if they are not precise. Of course, researchers should still pursue accuracy, but if current trends hold research will make little impact on policy or public perception in the coming years.
I laughed so hard I cried during the TikTok commentary. SO funny. Tysm for bringing joy and laughter to a tough week 💖
Have we considered using human beings as batteries for AI? (Not a bot, but asking for a friend who is a bot.)
Hard Fork, have you paid your cat tax?
To female employees at Meta - your thoughts on Zuck's comments? Or is the money coming in so good that you (just like white women this past election) look the other way and justify/excuse his words?
Hello youtube
Please stop the non-stop gratuitous silly-girl laughing. Sooo Annoying.
Try to say "hard fork" in youtube search, what do you get?
I loved this podcast until today. When people start to misrepresent dialogs of people they don't like on purpose and clearly show an excerpt of the dialog that has nothing to do with what they are actually inferring, that's when any person with principles has to move away. Thank you for your tech related content until now but I'm unsubscribing today.
Bye! 😘
Adiós! 😘
Listen again to what they say at 1:11 (at the end)