Orbit is an LLM addon/extension for Firefox that runs on the Mistral 7B model. It can summarize a given webpage, YouTube videos and so on. You can ask it questions about stuff that’s on the page. It is very privacy friendly and does not require any account to sign up.
I personally tried it, and found it to be incredibly useful! I think this is going to be one of my long term addons along with uBlock Origin, Decentraleyes and so on. I would highly recommend checking this out!
The general tone in this thread seems so very different from when “Mozilla is working on AI” was first announced
I’m just glad it’s an add on/extension. A lot of the crap baked into browsers these days is just bloat nobody wants or uses.
Most important part of the thread:
In it’s beta stage, Orbit is currently not open-source. This doesn’t mean it will remain this way forever. If orbit gains traction and we have the resources and funding to support an Open-Source project, I’m sure things could change.
Press X to doubt.
Has Mozilla done sometime to deserve this skepticism? They were founded on open-source and AFAIK have continued to support open-source. Mozilla is far from a perfect organization, but if this project was a success I think it would be out of character for them to keep it closed-source.
Eh, skepticism should be the default.
But I agree with you, nothing they’ve done is inherently bad, though they’ve done some abysmally stupid things in the way they handle them.
But I also really wish they’d stop fucking around with half-assed things like this and focus on core utilities.
then why make it closed source to begin with?
Believe it or not but it requires resources to open source an internal product, especially one that may have been an experiment where some small team was able to convince leadership could become useful to the masses.
React.js at Facebook is a good example of this. It took a lot of effort to externalize and open source React, and tbh the codebase is still kind of garbage when it comes to contributions from those unfamiliar with its intricacies.
but… you dont have to accept contributions? you can just make it open source and tidy it up at the same time?
So risk someone else beating you to market? And they’ll either have the resources to make it superior, therefore making yours irrelevant, or they’ll make it inferior, which generated bad press for you
In a different world maybe, but I can already see the headlines, “Mozilla open sources lackluster AI tool”. PR is unfortunately a thing, and once you miss that initial wave of interest, you’re unlikely to grab attention later without another marketing push. Mozilla is experienced in open sourcing software, so by now they’re pretty good at knowing when to do it and when not to. In other words, it says something that they chose not to do it in this case.
Yeah, it definitly tells me something, namely that I should not use the tool.
Why would news publish articles about the code quality of the tool, instead of its functionality?
Now they have negative press about its closed source nature, which is a calculated risk they took, just to open source it soon anyway? I doubt it.
It’s provocative it gets the people going.
Pocket.
That’s a pretty good answer. I knew Mozilla had bought it, and were operating it as an independent subsidiary. I didn’t know they promised to open-source it over 7 years ago.
Firefox is sustained (biggest funder) by google who needs artificial competitions to not be labeled a monopoly.
Its still the best browser i can think off that isn’t chromium but i would recommend staying skeptical.
Ooh, I just tried it out and I can tell I’m going to love it - if not this specific plugin (the UI needs some work) then this general concept of a plugin.
I just popped over to Youtube and went to a ten-minute video of something or other, clicked the “summarize transcript” button, and within a few seconds I had a paragraph-long summary of what the whole video was about. There have been sooo many Youtube videos over the years that I’ve reluctantly watched with a constant “get to the point, man!” Frustration. Now I’ll know if it’s worth it.
Hm…could be useful for those times you want to read a guide but can only find one in video form
This sounds like a good concept.
RIGHT?!!! IT’S SO FKIN AMAZING
This is especially going to be useful for me as a student. It’s just feels like browser 2.0 at this point haha
Probably not for me as I’m not interested in a summarizing tool, but I’m not against AI in general.
OAN, I think over time, the community will see that AI was a bubble, but in the same way that the internet was a bubble back in the day.
OAN, I think over time, the community will see that AI was a bubble, but in the same way that the internet was a bubble back in the day.
Surprised to see this opinion on Lemmy haha. Yep, totally agree with ya here!
Everyone wants a Her style personal assistant — as in one that is personal-context aware, can simplify, and generally enrich their lives (not for emotional support) — but if most people knew how unintelligent AI is, how spectacularly it fails, and how dangerous it is to integrate it into information systems and (especially) give it any ability to act … Literally nobody would want to give it access to all their data, or use it beyond an advisory role.
In before is not just skips important details in its summarization, but also hallucinates its own interpretation of things into it.
Generally, don’t call it “AI”, don’t overhype it, don’t use it where it is bad in its function (like telling you “facts”), don’t shove it into everything. I bet 80+ percent of all “AI” energy consumption is wasted on completely useless and moronic tasks that have 0 value even on a personal level.
"The term “AI” has been in use since 1956 for a wide range of computer science techniques. LLMs most certainly qualify as AI. You may be thinking of the science-fiction kind of “artificial people” AI, which is a subset of AI called Artificial General Intelligence when researchers want to be specific about that kind.
I’m thinking of something that actually processes some form of “thought”, in the abstract sense. Even video game AI does that to an extend (granted, there’s various techniques depending on the game type), so the term here is actually somewhat appropriate. LLMs don’t do that at all though, they’re just word guessing based on the texts they were trained upon (while we stick with text gen here at least) and that just so happens to sound like somewhat coherent sentences that can fool someone into thinking that their computer actually talked to them. There never was any sort of thought behind that though. It functions closer to how your mobile keyboard predicts the next word you want to use in its suggestions at the top. It just tries to complete the text it was already presented with. A lot of the illusion here comes actually from the tools used to display this information in a chat like manner, but that’s just frontend foolery for the user.
I think it’s more that you’re overestimating video game AI, here. If your definition of “abstract thought” doesn’t include what LLMs do then it definitely shouldn’t include video game AI. It’s even more illusory.
Yeah but you think a lot of weird things, so that does not surprise me in the slightest.
I would agree with the other guy. A video game AI can be as simple as some if-then decision logic, and i would count that as “AI”. An LLM also makes “decisions” on what to do/say, just via a different mechanism (predictive modeling) . I would still bucket that as AI. It you count one you should count the other. Neither are truly “thinking” in the sense of an AGI.
Gross.
If you really care for an LLM, run it locally… Not sure if this does it…
Don’t want to install and maintain 10gigs of cuda stuff on my PC. Next, my mum won’t know how to do that. Her laptop is a potato. This add-on makes all of this way easier.
You don’t need CUDA, it’s actually pretty easy. You can run the Mistral 7B model this add-on is based on using GPT4All. It doesn’t require much, if any, technical knowledge.
HOLY HELL THAT’S COOL. It can do so much too!!!
I locally installed some small LLM model more than a year ago. It took up like 25 gigs or something along with all CUDA libraries n stuff. It was alright, but I figured that cloud based solutions were the best for my use case, as they were better and for free.
I had no idea that open sourced AI progressed so much in the last year. Amazing stuff!
You’re not generating models at this point. You don’t need that kind of hardware to run these.
Well that comes with shit ton of privacy risk. If y’all are comfortable, then it is your choice
