@feditips building the language model is what takes a lot of resources. Running the model is pretty minimal and often can be done totally locally on something as low-powered as a cell phone.

I’d say this is an incorrect take. 🤷‍♂️

@feditips I'm a little confused, should we also not be using speech to text or other voice recognition services because they all use ML these days? Or photo management stuff like organizing by face and search for types of photos that have been processed by ML?

Of course that stuff is increasingly moving to be on device, so maybe that is cheap enough? Still uses power though.

I also honestly don't know the cost of image to text vs. the image generation talked about in the article.

@feditips carbon is just plant food...

@feditips can also recommend to run ML models locally on your laptop with ollama or gpt4all.

@feditips

Speaking of descriptions, I want to encourage everyone to provide descriptions for links that are not self-descriptive. Youtube is a perfect example.

I am less likely to chase a link if I do not know what the subject matter is.

@feditips definitely agree with you on that , it should be reserved to people with difficulties

@feditips bruh

@feditips Just thinking: how much CO2 do we produce for publishing 1 toot? For real.

@feditips
I agree with this, although I doubt I would have brought up the environment argument.