I could say it's equally close minded not to sympathize with this position, or various reasoning behind it. For me, I feel that my spoken language is effected by those I interact with, and the more exposed someone is to a bot, the more they will speak like that bot, and I don't want my language to be pulled towards the average redditor, so I choose not to interact with LLMs (I still use them for code generation, but I wouldn't if I used code for self expression. I just refuse to have a back and forth conversation on any topic. It's like that family that tried raising a chimp alongside a baby. The chimp did pick up some human like behavior, but the baby human adapted to chimp like behavior much faster, so they abandoned the experiment.)
I try to be polite just to not gain bad habits. But, for example, chatGPT is extremely confident, often wrong, and very weasely about it, so it can be hard to be “nice” to it (especially knowing that under the hood it has no feelings). It can be annoying when you bounce the third idea off the thing and it confidently replies with wrong instructions.
Anyway, I’ve been less worried about running local models, mostly just because I’m running them CPU-only. The capacity is just so limited, they don’t enter the uncanny valley where they can become truly annoying.
I do also find that only using a turn signal when others are around is a good reinforcement to always be aware of my surroundings. I feel like a jerk when I don't use one and realize there was someone in the area, just as I feel like a jerk when I realize I didn't turn off my brights for an approaching car at night. In both cases, feeling like a jerk reminds me to pay more attention while driving.
Signalling your turns is zero cost, there is no reason to optimize this.
That is a very bad habit and you should change it.
You are not only signalling to other cars. You are also signalling to other road users: motorbikes, bicycles, pedestrians.
Your signal is more important to the other road users you are less likely to see.
Always ALWAYS indicate. Even if it's 3AM on an empty road 200 miles from the nearest human that you know of. Do it anyway. You are not doing it to other cars. You are doing it to the world in general.
Here is a hypothetical: A loved one is being hauled away in an ambulance and it is a bad scenario. And you're going to follow them. Your mind is busy with the stress, trying to keep things cool while under pressure. What hospital are they going to, again? Do you have a list of prescriptions? Are they going to make it to the hospital? You're under a mental load, here.
The last thing you need is to ask "did I use my turn signal" as you merge lanes. If you do it automatically, without exception, chances are good your mental muscle memory will kick in and just do it.
But if it isn't a learned innate behavior, you may forget to while driving and cause an accident. Simply because the habit isn't there.
It's similar for talking to bots, as well. How you treat an object, a thing seen as lesser, could become how a person treats people they view as lesser, such as wait staff, for example. If I am unerring polite to a machine with no feelings, I'm more likely to be just as polite to people in customer service jobs. Because it is innate:
Watch your thoughts, they become words; Watch your words, they become actions.
This has a failure state of "when there's a nearby car [or, more realistically, cyclist / pedestrian] of which I am not aware". Knowing myself to be fallible, I always use my turn signals.
I do take your point about turn signals being a reminder to be aware. That's good, but could also work while, you know, still using them, just in case.
I have no opinion on not wanting to converse with a machine, that is a perfectly valid preference. I am referring more to the blog post's position where it seems to advocate against itself.
It's mostly knee-jerk reaction from having AI forced upon us from every direction, not just the ones that make sense
(It's weird how people can be so anti-anti-AI, but then when someone takes a middle position, suddenly that's wrong too.)
The focused purpose, I think, gives it more of a "purpose built tool" feel over "a chatbot that might be better at some tasks than others" generic entity. There's no fake persona to interact with, just an algorithm with data in and out.
The latter portion is less a technical and more an emotional nuance, to be sure, but it's closer to how I prefer to interact with computers, so I guess it kinda works on me... If that were the limit of how they added AI to the browser.
> Large language models are something else entirely. They are black boxes. You cannot audit them. You cannot truly understand what they do with your data. You cannot verify their behaviour. And Mozilla wants to put them at the heart of the browser and that doesn’t sit well.
Like I said, I'm all for local models for the exact reasons you mentioned. I also love the auditability. It strikes me as strange that the blog post would write off the architecture as the problem instead of the fact that it's not local.
The part that doesn't sit well to me is that Mozilla wants to egress data. It being an LLM I really don't care.
Not everyone uses their browser just to surf social media, some people use it for creating things, log in to walled gardens to work creatively. They do not want to send this data to an AI company to train on, to make themselves redundant.
Discussing the inner workings of an AI isn't helping, this is not what most people really worry about. Most people don't know how any of it works but they do notice that people get fired because the AI can do their job.
A local model will have fewer filters applied to the output, but I can still only evaluate the input/output pairs.
It’s insane this has to be pointed out to you but here we go.
Hammers are the best, they can drive nails, break down walls and serve as a weapon. From now on the military will, plumber to paratrooper, use nothing but hammers because their combined experience of using hammers will enable us to make better hammers for them to do their tasks with.
Firefox could have an entire section dedicated to torturing digital puppies built into the platform and... Ok, well, that's too far, but they could have a costco warehouse full of AI crap and I wouldn't mind at all as long as it was off by default and preferably not even downloaded to the system unless I went in and chose to download it.
I know respecting user preference doesn't line their pockets but neither does chasing users down and shoving services they never asked for and explicitly do not want into their faces.
An ideal translation is one which round-trips to the same content, which at least implies a consistency of representation.
No such example or even test as far as I know exists for any of the summary or search AIs since they expressly lose data in processing (I suppose you could construct multiple texts with the same meanings and see if they summarize equivalently - but it's certainly far harder to prove anything).
It's not a lossy process, and N round-trips should not lose any net meaning either.
This isn't a possible test with many other applications.
Translation is lossy. Good translation minimizes it without sounding awkward, but that doesn't mean some detail wasn't lost.
To me the difference between something like AI translation and an LLM is that the former is a useful feature and the latter is an annoyance. I want to be able to translate text across languages in my web browser. I don't want a chat bot for my web browser. I don't want a virtual secretary - and even if I did, I wouldn't want it limited to the confines of my web browser.
It's not about whether there is machine learning, LLMs, or any kind of "AI" involved. It's about whether the feature is actually useful. I'm sick of AI non-features getting shoved in my face, begging for my attention.
This really weakens the point of the post. It strikes me as a: we just don't like those AIs. Bergamot's model's behavior is no more or less auditable or a black box than an LLM's behavior. If you really want to go dig into a Llama 7B model, you definitely can. Even Bergamot's underlying model has an option to be transformer-based: https://marian-nmt.github.io/docs/
The premise of non-corporate AI is respectable but I don't understand the hate for LLMs. Local inference is laudable, but being close-minded about solutions is not interesting.