9mo ago

Privacy focused ChatGPT in you terminal

so, about a month ago i stumbled upon the duckduckgo ai chat feature and wrote an article about how private their APIs are, and a few weeks after, a CLI client.

the thing is in a pretty mature stage now (its not like there is a lot of work to be done there tbh)

its not super private, but it is "private enough". the only thing duckduckgo has is your IP, which is usually not much unless you are on a residential connection with a dedicated IP

17 comments

Just a PSA to anyone who looks here: with DDG, ChatGPT is not your only option. You can choose to use Claude instead, if you wish.
- And recently Llama3 and Mixtral!
  
  I actually didn't know that! Thanks!
  
  Llama3 isn't free

What's privacy-focused ChatGPT? Is it like diet butter?
Hint: if it doesn't run on your machine, it's collecting monetizable data.

After enjoying the very local power of Ollama it seems weird to give OpenAI any money.
- Mistral is my favorite model
  
  Mine too
  Looks like Mistral was trained more on proven academic data and less on Facebook posts. This does make it a bit less variable in answers, but at least it says very little bullshit.

I don’t need artificial intelligence in my terminal. Do you know how many times some troll has posted about “rm -fr /” on Reddit and other shitty forums, which then gets gobbled up and laundered by LLMs? Not letting that anywhere near my prod servers with valuable data.
- Do you know how many times some troll has posted about “rm -fr /” on Reddit and other shitty forums, which then gets gobbled up and laundered by LLMs?
  Ironically, that now includes you 🙂

This is about LLMs, not Linux.

Disclaimer: The following section is of educational purposes only. Please do not scrape DuckDuckGo's APIs, as it would only hurt the company which honestly cares about your privacy.
😆

Thank you this looks interesting!

Can we keep LLM BS in "AI"-specific subs, please?
- I don’t get why you’re getting downvoted. You’re right. The topic is too focused on LLMs, and is not generally applicable, or exclusive, to Linux.
  
  I've run across AI content in non-AI subs before and responded with similar things fully expecting to be downvoted to Earth's core and was pleasantly surprised at how few downvotes and how many upvotes I got. So I've made a habit of calling out AI stuff in non-AI communities when I run across it. (Unless it's actually strongly related to the purpose of the community it's posted in.)
  This is at least the third time I've made such a comment, and the first time I've seen a negative net score for more than a few minutes after I posted. Still, it's pretty evenly split even on this one. Maybe most of the down-voters just don't want anything negative said about "AI". (Like cryptobros will deem anything "bearish" to be "FUD".) Who knows.
  And there are places for engaging in mouth-foaming AI bubble hype indulgence on Lemmy. I just don't want the communities I like inundated with tons of AI PR posts.

17 comments