Misread the summary to say data driven delusions and that seems accurate.
this post was submitted on 13 Aug 2024
-33 points (24.6% liked)
Technology
59135 readers
2842 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
Put my entire dataset into an LLM so whoever developed the LLM can steal it and use it for training? No. Thank. You.
massive security / data protection issue.
Both langchain as well as ollama run locally and are open source.
To be very frank: your post sounds like fear mongering without having even clicked on the link.
I did read the article. Do you arbitrarily trust any code you run locally without reviewing it?
I have reviewed the tiniest fraction of code that I have ever used.
$ dpkg -l|wc -l
4526
$
That's about 4500 software packages I have installed on one Linux system, to say nothing of other computing devices I've used or the other packaging systems in use on this system alone. I have probably looked at any portion of...I don't know, maybe 20 of those? And that's to work on a small portion of any one's codebase, certainly not to audit the software package.
Nobody using any kind of a remotely normal and modern computing environment, even if they are a software developer and know at least one programming language used by some of the software on their system and if they have the relevant domain knowledge to assess security concerns, has the realistic ability to conduct a review of the code that runs on their system, even in environments, like Linux, where the code is available.
It's like asking a mechanical engineer to validate the design correctness of every mechanical device they've ever used prior to using it.
But I mentioned llama, which is self hosted
Just learn the syntax? It's not super complicated, and the whole reason we have strict syntax is because you can't reliably convey intent without it.
Or build a better front end.
And their directions for setting this up are vastly more complicated than a basic SQL query.
There's almost nothing I want less than chatting with my database(s). These people are desperate.
I mean you kinda of looking at from the wrong way, In the company I work for, it was a time saver. A department of engineers in embedded systems can upload some measurements then another department can quicky run some questions to write the findings for a customer. They both don't know anything about SQL and ti is even faster than the tool we built for them. Since they have to filter through 1TB of data. Yes it needs a lot of optimization to get it perfect.