Assuming you meant harm. My assumption is that this is a locally CPU run model that is basically a glorified AOL chatbot. Again, I’m making some assumptions here.
But I really think the “AI” in this is just using a buzzword. When technically it is but it’s likely not what we think like a heavy LLM like ChatGPT. I really doubt Firefox is sending prompts to GPU farms and it’s likely just a simple model that runs on your local CPU. Likely optimized to answer browser related prompts or Firefox related prompts in general. I could be wrong. But I’d be surprised if it was at all outsourcing prompts to a GPU farm.
It’s a good example (if my assumptions are correct and Firefox is not braindead plugging in ChatGPT into the browser) actually of how AI application specific models can be very efficient and helpful. All the money we push into the “god AI” is really what people think of with “fuck AI”. China is working heavily in models like this.
For example. They just released a new application specific model that is extremely efficient at character recognition in images. Basically, if you took a picture of a page in a book it converts it to text that can be edited on a computer. It’s a case of an AI model actually improving compute efficiency when compared to other previous “non-ai” methods of computer vision.
But I digress. I think most of “fuck AI” could be summarized as “fuck capitalism” though. American investors are chasing a “worker replacement” model that they are failing at obtaining.
Completely get your view there, like whenever I’m talking to people about tech (especially when I’m having other scientists explain their research) I always avoid using the word AI and when they use it I always tell them to clarify.
I have no idea what they are using for their own summarization features, but I know the chatbot tool is just a front-end to interface with other services, not its own LLM.
Also which model are you talking about for image to text, sounds like a potential improvement on earlier work I’d like to see
I do agree as well most “fuck AI” is fuck capitalism if only indirectly by some. Its why I care about the environmental harm a lot as it wouldn’t exist if not for the capitalistic drive to build at all cost. LLMs aren’t worth people’s lives or dead ecosystems, they aren’t even the tech that will achieve AGI as they don’t do continuous learning (evident from the seed change between shit GPT3 to human like ChatGPT was an RL training protocol). Most modern technology has the curse of its owners which prevents it from being used ethically or for the common good.
Don’t forgot environmental hard
Assuming you meant harm. My assumption is that this is a locally CPU run model that is basically a glorified AOL chatbot. Again, I’m making some assumptions here.
But I really think the “AI” in this is just using a buzzword. When technically it is but it’s likely not what we think like a heavy LLM like ChatGPT. I really doubt Firefox is sending prompts to GPU farms and it’s likely just a simple model that runs on your local CPU. Likely optimized to answer browser related prompts or Firefox related prompts in general. I could be wrong. But I’d be surprised if it was at all outsourcing prompts to a GPU farm.
It’s a good example (if my assumptions are correct and Firefox is not braindead plugging in ChatGPT into the browser) actually of how AI application specific models can be very efficient and helpful. All the money we push into the “god AI” is really what people think of with “fuck AI”. China is working heavily in models like this.
For example. They just released a new application specific model that is extremely efficient at character recognition in images. Basically, if you took a picture of a page in a book it converts it to text that can be edited on a computer. It’s a case of an AI model actually improving compute efficiency when compared to other previous “non-ai” methods of computer vision.
But I digress. I think most of “fuck AI” could be summarized as “fuck capitalism” though. American investors are chasing a “worker replacement” model that they are failing at obtaining.
I did mean harm yeah
Completely get your view there, like whenever I’m talking to people about tech (especially when I’m having other scientists explain their research) I always avoid using the word AI and when they use it I always tell them to clarify.
I have no idea what they are using for their own summarization features, but I know the chatbot tool is just a front-end to interface with other services, not its own LLM.
Also which model are you talking about for image to text, sounds like a potential improvement on earlier work I’d like to see
I do agree as well most “fuck AI” is fuck capitalism if only indirectly by some. Its why I care about the environmental harm a lot as it wouldn’t exist if not for the capitalistic drive to build at all cost. LLMs aren’t worth people’s lives or dead ecosystems, they aren’t even the tech that will achieve AGI as they don’t do continuous learning (evident from the seed change between shit GPT3 to human like ChatGPT was an RL training protocol). Most modern technology has the curse of its owners which prevents it from being used ethically or for the common good.