AI assistant in a browser...it's getting ridiculous.
Ridiculous maybe, but by no means pointless. “AI” will be the end of ad blockers and a whole host of measures to secure a remnant of privacy on the net.
On the contrary, I feel like AI--even if it doesn't advance much further than the level we already have achieved--is putting us at the cusp of finally being able to have good end-game fully-working ad blockers: I'll just have an AI in my browser read / look at / watch / "experience" your content and then launder away any and all ads in its presentation to me... I can even have it make extreme edits if required, rebuilding the audio and video, to remove subtle bias your content wants to infect me with due to sponsorship deals.
What is fighting to prevent ad blockers is not AI... it is pervasive DRM: if we live in a world where the tech platforms continue to side with Big Content and prevent us from being able to capture and alter media as we see it on devices they insist we "buy" and yet never "own"--and where the business models of the most powerful companies rely on proving that an end-user human affirmatively was on the other side of an interaction (and thereby might have been infected with paid propaganda)--we are stuck inside of a dystopia :(.
[1]: https://www.eff.org/deeplinks/2024/02/dont-fall-latest-chang...
If only. This wave of AI integrations is primarily focused on pleasing (potential) investors.
This wave of AI integrations is happening in a context/framework for normalizing naked avarice so that preposterous power can be concentrated in the hands of the rich and dubiously ethical.
for example, I don't want to see yuppie shit on news.ycombinator.com. right now, I have a userscript that filters out links from the worst offenders (e.g. newyorker), but every day, there are many more that shit up the front page and waste my time and attention. this could be solved with a very low parameter LLM by asking it to evaluate whether the text is related to science and technology or not.
hell, even images could be filtered with vision models, including local ones. I'd fucking love to hide a few broad categories of images from my sight, e.g. clickbaity youtube thumbnails (DeArrow helps to a degree)
FWIW, I once used naive Bayes classification solely on HN headlines and it worked reasonably well for solving the problem you're describing.
I summarize the transcripts of 20+ min. long youtube videos to get the gist, which is what I am interested in anyways. Saves me time. (this is something I built custom for my use, maybe there are other options for this)
It is not an obsession. I always wanted something, someone that could do this for me, and now we have it.
If the AI is summarizing files the user downloads, why couldn't this be done with external tools?
In general though, why does the browser itself need to be "smart"? Its primary, and perhaps only, goal should be to render web pages... I feel like this push to integrate AI features into browsers is done so that companies can promote their own (completely unrelated) AI services, and pull people even more into their ecosystem. It's obnoxious.
Most AI tools I've seen however can't seem to get away from the classic "prompt -> response" chat UI.
I can't say I will use every ai feature, but so far they've been helpful. I don't feel more ignorant. What I think makes people ignorant is outsourcing labor tasks like a cleaner for the home, a gardener for the yard, or handyman for basic repairs. I don't think those are necessarily wrong to outsource though.
Anyway, now that this sort of technology exists, Microsoft is missing a great opportunity to bring back clippy and cash in on some nostalgia. I miss that lil guy.
It's pretty interesting! They do basically the same thing for core Chromium, applying a (big) set of patches[1].
Incidentally, I'd be interested to hear any ideas/approaches to this problem. I'm guessing if there was something clearly better, Brave would be doing it, but it seems like there should be a better way even if I can't think of one.
What you'll notice more often is a folder we have called `chromium_src`. This directory mirrors the directory structure for Chromium under `src` and the build system will look for matches. If there's a file with the same name under `chromium_src`, it'll prefer that one. That file then does what it needs to differently and then includes the original file.
This approach helps keep things much more lightweight - but it has challenges too. If code fails to apply (file that `chromium_src` is matching gets renamed, etc) it can be hard to detect. This is where you'd want to have a test to catch that.
Another person shared - but here's a link to our patching documentation: https://github.com/brave/brave-browser/wiki/Patching-Chromiu...
You'll notice the actual patching itself is introduced with the caveat:
> When other options are exhausted, you can patch the code directlyThey document how developers should "rebase" chromium:
https://github.com/brave/brave-browser/wiki/Chromium-rebases...
And it looks like way more work than doing it with git via
git rebase chromium/masteri have rafts of bookmarked pages, but i don’t know which of the 45 postgres-related pages contains a helpful sentence or phrase about indexing that i think i remember reading a few months ago.
Ollama /JUST/ recently added some (fast) embeddings models: https://ollama.com/library/nomic-embed-text
(note: not OP)
open source AI and browser/OS companies are going to provide an answer soon anyway.
That's when I quit.
(It was a simple sheet. I said "list the names of all items where [column name] is false". There was about 60 rows)
So according to Brave, using their AI leaks:
- your searches
- What you're viewing
- What you're typing
Did Brave just turn into the Chrome that it once so hated? I guess it is just orange chrome.
The only issue I have is that the sync between my iphone, macbook, windows can sometimes get out of sync.
Yes it has Tor built-in, but you don't ever need to use it - it's just there in case you do (I have never needed it myself).
I don't think I'm part of a cult, but I've used FF as my default browser for over a decade and I guess I don't know what else people want from a browser.
I guess I'm in this cult. But one of the big reasons is what else is fighting the chromium monopoly? Safari? Just imo there's not too big of differences between browsers and people just exaggerate these differences.
I know what you're gonna say, but forgetting everything else there's one important factor you should consider. Chrome has a (near) monopoly and resolving that monopoly requires using non-chromium browsers.
But on top of that, Firefox is fast, secure, has privacy in mind, and a rich set of add-ons. But most of that is true for any browser you pick. There really aren't big differences between browsers and often we're making mountains out of mole hills when we compare. But I'll say, firefox has ad-blocking on mobile (plugins on mobile, like 800 exist)
Haven't had Chrome installed on any of my machines for ~5 years at this point.
Not the biggest fan of Brave (especially considering this latest AI crap, and all the weird crypto stuff), but I'm satisfied with it overall. FF still remains #1 in my eyes and usage, but has to be tweaked to my liking.
A tip I found recently in about:config
browser.compact.show=true to bring back the compact layout option, results in very good use of space on laptop along with vertical tabs and also Firefox allows the vertical tabs to be moved to right side which is nice.
I work for Mozilla (speaking for myself, of course), so I'll leave you to guess which I'd recommend :P
[1] https://github.com/gorhill/uBlock/wiki/uBlock-Origin-works-b...
[2] https://arstechnica.com/gadgets/2023/09/googles-widely-oppos...
I haven't had issues with uBlock Origin - very occasionally an ad will seep through on YouTube, maybe once every 6 months, but when that happens I refresh the page and the ad is gone.
Safari is the new IE and Firefox I've always found to just be alright - for sure not as big of a fan of Firefox dev tools over Chrome dev tools. And Firefox scrolling behavior can be annoying.
For all personal browsing and projects, I generally used Firefox, but switched over a couple years ago to using Brave on the phone, and am kind of half-transitioned from Firefox to Brave on desktop. I've been a Firefox user forever, but it's slowly losing me.
Cromite has a desktop build, but it's a bit more experimental than the mobile build, so you can use Ungoogled Chromium[1] instead. Ungoogled is also a privacy-oriented open source patchset on top of Chromium. Check the beta flags to enable some more interesting features like getClientRect anti-fingerprinting measures (unfortunately breaks some React-based sites that go into infinite re-render loop).
Both of these browsers selectively include patches from Brave, but they are community-oriented builds so imo more trustworthy than Brave, which continues to package various shady anti-features and always will because it's backed by a for-profit company.
LibreWolf[2] is the nicest Firefox-based one for desktop, I think. It's pretty hardcore, though, I most only use it to visit mainstream social media sites.
I tried a bunch of the Firefox-based ones on mobile and none of them clicked for me. Cromite is just too slick on Android. Put the address bar at the bottom and off you go. Only downside is no online syncing of tabs and bookmarks, but meh. You can save all open tabs to bookmark bar in one hit then export your bookmarks, send the file through whatever E2EE channel you want to your other device, import and reopen them again.
[0] https://github.com/uazo/cromite
[1] https://github.com/ungoogled-software/ungoogled-chromium
I tried using Firefox on and off, but it's sadly technically inferior to Chromium, and that gap has gotten worse. It still has things that it does better than Chromium browsers, like history sync actually working, or reader view. But those are few and far between. And Multi-Account Containers, one of its apparent advantages, comes with the caveat that Firefox doesn't have usable profiles and the security for its extensions is worse (e.g., no click to activate, no ability to disable extensions in certain containers).
What finally pushed me to Chromium is the poor PWA support in Firefox. On Android, it has bugs that haven't been fixed for years (never mind the poor performance that's well known), and on desktop they've basically dropped the ball.
I use several PWAs. If it's a chat app, I use it as a PWA. Also Spotify, since you mentioned it ... as I like having better sandboxing and ad-blocking in my apps. On Android, too.
Also I think they have loads of funding, and are factoring all of this into user acquisition costs
Yes, it defaults to enabling their advertising features, but that's just it, a default setting, it isn't hard to disable if you don't want to "take part of the experiment".
It does not, you have to opt in.
You are right in the rest of your comment thought. And in general, when you compare default configs, Brave does far more to protect your privacy than Firefox does.
Well maybe due to the multiples bad things that brave did in the past.
Eich bad