This thought entered my mind today as I came across a thread on Quora, and noticed that they have added a feature where ChatGPT would have a go at answering the question.
Today alone I have used a few varying “AI” tools, including one which automatically paraphrases text for you, one which analyses your writing in SwiftKey as you type, and of course the big players like Bard and Bing Chat. It got me thinking about whether these features are actually valuable, and if we would start to see them on this platform.
No. I come here to interact with people. One of my favorite features here is the setting that lets me hide all bot accounts.
Wh… Why? What do you imagine it adding to a social space?
What do you think it adds to a keyboard (swiftkey)? The feature is there regardless whether we like it or not. The question is whether you think such a thing should be on the site, and in what form, not whether one user considers to be useful or not.
One thing that it could do is analyse your perceived tone, and suggest edits to come across less aggressive, or more inquisitive for example. Such a thing could well make certain social places a more pleasant forum to use.
One thing that it has added to keyboards is making it much more difficult to use purposefully creative spelling or grammar.
For every problem they fix, they restrain creativity, because they themselves are constrained to their training set.
The Internet already has a problem with tone policing. Maybe we don’t need mathematical models burning the planet to tell me that in pissed off.
If people want a moderated space, they can request some moderators.
Also, you didn’t answer my question. Examples of what cna be done are not the same as answering why a thing should be done.
Nah, I’m good.
I don’t think so. Integrating with existing AI services equals direct privacy concerns to me.
A solution would be a self-hosted model, but then that could cost a lot in machine resources for instance owners, especially bigger instances.Even then I’m having a hard time thinking of a useful usecase for an LLM integration on Lemmy (or even social media in general).
Nope. AI is all well and good for Quora and Reddit, but it can keep its nose out of the fediverse.
Which ML model, is it proprietary? which service would handle the requests? which servers would do the computation? etc
Most here are seeking freedom away from the corporate tech monsters, not to give them a seat at the table.
The massive push towards everything integrating AI legitimately scares me. These systems have zero motivation to post things that are factually accurate or unbiased, and yet people are happy forcing them everywhere.
On a social media site, I’d like to interact with genuine people with genuine thoughts and experiences, not a faceless conglomerate of a bunch of books and reddit posts.
And of course, there’s the ethical and privacy issues of these systems just scraping and regurgitating huge amounts of content without regards to whether the original creators are okay with it or not.
No.
It’s really odd the take some people have with “AI” dislike the term since it’s too broad. If anyone here has a high end modern smartphone they have dedicated chips to ML, predictive text, virtual assistants etc. odds are most people here use a device daily that uses “AI” or have the lot of you gone down to older smartphones or use non smart devices?
It’s not odd, we’re just able to read for understanding. No one is asking if we want colour correction in photos here, just like no one is asking if we want linear regressions or MCMC models.
They’re asking do we want to include generative LLMs.
I think the crux of the issue is that corporations are quick to throw around the term “AI” as it’s a buzz word, and the lay person does not know what it means other than “smart-ish”. I’d argue that there is no real AI in existence (yet).
