9 ms·

The title does not match the content. A more appropriate title is "Emissions caused by chatgpt use are not significant in comparison to everything else." But,

The title does not match the content. A more appropriate title is "Emissions caused by chatgpt use are not significant in comparison to everything else." But, given that title, it becomes somewhat obvious that the article itself doesn't need to exist.

jonas21 2y ago
> given that title, it becomes somewhat obvious that the article itself doesn't need to exist. Why? I regularly hear people trying to argue that LLMs are an environmental distaster.
- _ache_ 2y ago
  Because LLMs are an environmental disaster. It's not about any individual usage. It's the global technology that is yet to prove to be useful and that already have bad for the environment. Any new usage should be free of impact on the environment. (Note: The technology of LLM itself is not an environmental disaster, but how it is put in use currently isn't the way).
  - satvikpendem 2y ago
    > It's the global technology that is yet to prove to be useful Useful for whom, by what definition? I personally find it very useful for my day to day work, whether it be helping me write code, think through ideas, or otherwise.
    - _ache_ 2y ago
      I don't want to be rude but I don't think that your personal experience is of any interest. The only question is "are LLM relevant ?" And the only way to assert that they are is to get numbers (big or small) and to compare them to alternatives.
      - satvikpendem 2y ago
        > I don't think that your personal experience is of any interest But you didn't specify what you meant by "useful," hence why I asked the question I did. So under such ambiguity, my assertion is absolutely a counterpoint to what you just said. I will ask again, useful for whom, by what definition?
        _ache_ 2y ago
        Oh! Good point, we may not have the same definition of usefulness. By useful I mean, to the world. That it affect the world in a good way. Maybe it's not the best technology to do something but replace the best way in an CO2 effective way. Maybe it is not a clean technology but increase overall fairness. I don't know, but it had to have a good income to the world in a way.
  - c0redump 2y ago
    > yet to prove to be useful I don’t understand this perspective. It should be abundantly clear at this point that these systems are quite useful for a variety of applications. Do they have problems? Sure. Do the AI boosters who breathlessly claim that the models are super intelligent make me cringe? Sure. But saying that they’re not useful is just downright crazy.
    - _ache_ 2y ago
      > I don’t understand this perspective. It should be abundantly clear at this point that these systems are quite useful for a variety of applications. LLM are polyvalent. But in most of the tasks they are not the most efficient way to do the task. Want to play chess ? Use Stockfish or Leela. Want to do image recognition ? SAM or TinyViT like models. Want to know if your are sick ? Go to the doctor or at least do a search on the web. Yes, there is tasks where LLM are perfect for (speech analysis/classification for example). But omnipotent chatbot isn't one for example. If there were a revolutionary use, we would have a productivity boom. We don't. This article is from 2021: https://www.technologyreview.com/2021/06/10/1026008/the-coming-productivity-boom/ https://www.technologyreview.com/2021/06/10/1026008/the-comi...
      - satvikpendem 2y ago
        > If there were a revolutionary use, we would have a productivity boom. We don't. What evidence do you have for this assertion? It seems like you are asserting something as fact when in reality it's your own personal opinion, yet ironically you are dismissing everyone else's personal experiences as mere opinion too.
        _ache_ 2y ago
        > What evidence do you have for this assertion? No predicted productivity boom (check last US data), no GDP boost yet (again last data). Even LLM enthusiast like McKinsey or Goldmansachs expect nothing before 2027. And it's not about LLM, it's about the whole AI progress. That is, obviously, a revolution. https://www.goldmansachs.com/insights/articles/ai-may-start-to-boost-us-gdp-in-2027 https://www.goldmansachs.com/insights/articles/ai-may-start-... But just to be clear. I'm denying something said to be obvious. I should not be the one who give sources about something that doesn't exist. If there is a productivity boom, I may not have seen it. Show it to me.
simonw 2y ago
The article needs to exist because the idea that ChatGPT usage is environmentally disastrous really has started to make its way into the human hive mind. I'm glad someone is trying to push back against that - I see it every day.
9rx 2y ago
> "Emissions caused by chatgpt use are not significant in comparison to everything else." Emissions directly caused by Average Joe using ChatGPT is not significant compared to everything else. 50,000 questions is a lot for an individual using ChatGPT casually, but nothing for the businesses using ChatGPT to crunch data. 50,000 "questions" will be lucky to get you through the hour. Those businesses aren't crunching data just for the sake of it. They are doing so ultimately because that very same aforementioned Average Joe is going to buy something that was produced out of that data crunching. It is the indirect use that raises the "ChatGPT is bad for the environment" alarm. At very least, we at least don't have a good handle on what the actual scale is. How many indirect "questions" am I asking ChatGPT daily?
deepsun 2y ago
Learning a new model (like GPT-4) is way more costly than running it.

jonas21 2y ago

> given that title, it becomes somewhat obvious that the article itself doesn't need to exist. Why? I regularly hear people trying to argue that LLMs are an environmental distaster.

_ache_ 2y ago
Because LLMs are an environmental disaster. It's not about any individual usage. It's the global technology that is yet to prove to be useful and that already have bad for the environment. Any new usage should be free of impact on the environment. (Note: The technology of LLM itself is not an environmental disaster, but how it is put in use currently isn't the way).
- satvikpendem 2y ago
  > It's the global technology that is yet to prove to be useful Useful for whom, by what definition? I personally find it very useful for my day to day work, whether it be helping me write code, think through ideas, or otherwise.
  - _ache_ 2y ago
    I don't want to be rude but I don't think that your personal experience is of any interest. The only question is "are LLM relevant ?" And the only way to assert that they are is to get numbers (big or small) and to compare them to alternatives.
    - satvikpendem 2y ago
      > I don't think that your personal experience is of any interest But you didn't specify what you meant by "useful," hence why I asked the question I did. So under such ambiguity, my assertion is absolutely a counterpoint to what you just said. I will ask again, useful for whom, by what definition?
      - _ache_ 2y ago
        Oh! Good point, we may not have the same definition of usefulness. By useful I mean, to the world. That it affect the world in a good way. Maybe it's not the best technology to do something but replace the best way in an CO2 effective way. Maybe it is not a clean technology but increase overall fairness. I don't know, but it had to have a good income to the world in a way.
- c0redump 2y ago
  > yet to prove to be useful I don’t understand this perspective. It should be abundantly clear at this point that these systems are quite useful for a variety of applications. Do they have problems? Sure. Do the AI boosters who breathlessly claim that the models are super intelligent make me cringe? Sure. But saying that they’re not useful is just downright crazy.
  - _ache_ 2y ago
    > I don’t understand this perspective. It should be abundantly clear at this point that these systems are quite useful for a variety of applications. LLM are polyvalent. But in most of the tasks they are not the most efficient way to do the task. Want to play chess ? Use Stockfish or Leela. Want to do image recognition ? SAM or TinyViT like models. Want to know if your are sick ? Go to the doctor or at least do a search on the web. Yes, there is tasks where LLM are perfect for (speech analysis/classification for example). But omnipotent chatbot isn't one for example. If there were a revolutionary use, we would have a productivity boom. We don't. This article is from 2021: https://www.technologyreview.com/2021/06/10/1026008/the-coming-productivity-boom/ https://www.technologyreview.com/2021/06/10/1026008/the-comi...
    - satvikpendem 2y ago
      > If there were a revolutionary use, we would have a productivity boom. We don't. What evidence do you have for this assertion? It seems like you are asserting something as fact when in reality it's your own personal opinion, yet ironically you are dismissing everyone else's personal experiences as mere opinion too.
      - _ache_ 2y ago
        > What evidence do you have for this assertion? No predicted productivity boom (check last US data), no GDP boost yet (again last data). Even LLM enthusiast like McKinsey or Goldmansachs expect nothing before 2027. And it's not about LLM, it's about the whole AI progress. That is, obviously, a revolution. https://www.goldmansachs.com/insights/articles/ai-may-start-to-boost-us-gdp-in-2027 https://www.goldmansachs.com/insights/articles/ai-may-start-... But just to be clear. I'm denying something said to be obvious. I should not be the one who give sources about something that doesn't exist. If there is a productivity boom, I may not have seen it. Show it to me.

simonw 2y ago

The article needs to exist because the idea that ChatGPT usage is environmentally disastrous really has started to make its way into the human hive mind. I'm glad someone is trying to push back against that - I see it every day.

9rx 2y ago

> "Emissions caused by chatgpt use are not significant in comparison to everything else." Emissions directly caused by Average Joe using ChatGPT is not significant compared to everything else. 50,000 questions is a lot for an individual using ChatGPT casually, but nothing for the businesses using ChatGPT to crunch data. 50,000 "questions" will be lucky to get you through the hour. Those businesses aren't crunching data just for the sake of it. They are doing so ultimately because that very same aforementioned Average Joe is going to buy something that was produced out of that data crunching. It is the indirect use that raises the "ChatGPT is bad for the environment" alarm. At very least, we at least don't have a good handle on what the actual scale is. How many indirect "questions" am I asking ChatGPT daily?

deepsun 2y ago

Learning a new model (like GPT-4) is way more costly than running it.