What? They’ve been talking about features that are now being called the “kill switch” for the better part of a year. Literally all they did that’s new was give it a dumb name.

No, why would they?

Serious and long answer because you won’t find people actually providing you one here: in theory (heavy emphasis on theory), an “agentic” world would be fucking awesome.

Agents

You know how you have been programmed that when you search something on Google, you need to be to terse and to the point? The worst you get is “Best Indian restaurants near me” but you don’t normally do more than that.

Well in reality most of the times when people just love rambling on or providing lots of additional info, so the natural language processing capabilities of LLMs are tremendously helpful. Like, what you actually want to do is “Best Indian restaurants near me but make sure it’s not more than 5km away and my chicken tikka plate doesn’t cost more than ₹400 and also I hope it’s near a train station so I can catch a train that will take me home by 11pm latest”. But you don’t put all that on fucking Google do ya?

“Agents” will use a protocol that works in completely in the background called Model Context Protocol (MCP). The idea is that you put all that information into an LLM (ideally speak into it because no one actually wants to type all that) and each service will have it’s own MCP server. Google will have one so it will narrow down your filters to one being near a train station and less than 5km away. Your restaurant will have one, your agent can automatically make a reservation for you. Your train operator will have one, so your agent can automatically book the train ticket for you. You don’t need to pull up each app individually, it will all happen in the background. And at most you will get a “confirm all the above?”. How cool is that?

Uses

So, what companies now want to do is leverage agents for everything, making use of NLP capabilities.

• Let’s say you maintain a spreadsheet or database of how your vehicle is maintained, what repairs you have done. Why do you want to manually type in each time? Just tell your agentic OS “hey add that I spent ₹5000 in replacing this car part at this location in my vehicle maintenance spreadsheet. Oh and also I filled in petrol on the way.” and boom your OS does it for you.

• You are want to add a new user to a Linux server. You just say “create a new user alice, add them to these local groups, and provide them sudo access as well. But also make sure they are forced to change their password every year”.

• You have accounts across 3 banks and you want to create a visualisation of your spendings? Maybe you want to also flag some anamolous spends? You tell your browser to fetch all that information and it will do that for you.

• You can tell your browser to track an item’s price and instantly buy it if it goes below a certain amount.

• Flying somewhere? Tell your browser to compare airline policies, maybe checkout their history of delays and cancellations

• And because it’s natural language, LLMs can easily ask to clarify something

Obvious downsides

So all this sounds awesome, but let’s get to why this will only work in theory unless there is a huge shift:

• LLMs still suck in terms of accuracy. Yes they are decent but still not at the level where it’s needed and still make stupid errors. Also currently they are not making as generational upgrades as before

• LLMs are not easy to self host. They are one of the genuine use cases of making use of cloud compute.

• This means they are going to be expensiveeeeee and also energy hogs

• Commercial companies actually want you to land on their servers. Yes its good that your OS will do it for you and they get a page hit but as of now that is absolutely not what companies want. How are they going to serve you ads?

• People will lose their technical touch if bots are doing all the work for them

• People do NOT want to trust a bot with a credit card. Amazon already tried that with Alexa/Echo devices and people just don’t like saying “buy me a roll of toilet paper” because most people want to see what the fuck is actually being bought. And even if they are okay, because LLMs are still imperfect, they are going to make mistakes now and then.

• There are going to be clashes of what the OS will do agentically vs what a browser will do. Agentic browser makers like Perplexity want you in their ecosystem but if Windows ships with that functionality out of the box then how much reason is there really to get Perplexity? I expect to see anti-competitive lawsuits around this in the future.

• This also means there is going to be a huge lock-in to Big Tech companies.

My personal view is that you will see some of these features 5-10 years down the line but it’s not going to materialise in the way some of these AI companies are dreaming it will.

Not entirely clear, but my best guess is that it will basically have an MCP implementation so that the browser can be controlled directly by an LLM

I think that’s basically what e.g. the chatgpt browser is. Despite the… hostile… response on the fediverse, I suspect it will end up being the way a lot of people interact with the internet in a few years.

The implementation challenge currently is that they’re extremely vulnerable to prompt injection.

Using existing LLMs functionality with fewer steps. You can have a chatbot in the side bar, no doubt keeping track of all your browsing habits to better assist you which incidentally builds a very valuable profile of the user companies would love to buy. Summarizing large texts so AI generated slop and search algorithm filler content can be filtered out more efficiently vs a decent chance at introducing errors. Rewording text so you can make it more simple, translated, adhering to your world view. All of this with minimal clicks, automatically done if possible.

Marketing

It will goon for you that’s what make the browser AI

The news of being able to use or disable all of the AI features was in the original announcement as well, but it was pretty clear that most of Lemmy just read the headline and leaned into it.

Firefox just can’t win with their users.

• Mozilla makes decisions based on market data
  • Users complain they never wanted those features
• Mozilla makes a decision based on user feedback
  • Users shit on them for backpedaling or damage control

It’s absurd.

For the record a quick web search for how to disable AI in firefox gave me this list of items to set to false in about:config :

browser.ml.enable
browser.ml.chat.enabled
browser.ml.chat.sidebar
browser.ml.chat.shortcuts
browser.ml.chat.page.footerBadge
browser.ml.chat.page.menuBadge
browser.ml.linkPreview.enabled
browser.tabs.groups.smart.enabled
extensions.ml.enabled

What is it actively doing now with AI? There is the ai sidebar, but if you don’t use that it isn’t used, right?

You reminded me that one use for AI I’d really like is removing all photos of Trump, Musk and Putin from my screen. Another is filtering the twenty reposts of every event in US politics and the incessant whining about prices. Alas, I need these in phone apps more than the browser.

Right right. If they had real innovation, they would have defined it clearly as you suggested. But they didn’t, so they don’t. It’s all snake oil, again, because that’s the entire AI industry.

• AI chatbot in sidebar (you can choose which chatbot you want, similar to how you choose default search engine)
• Shake to summarize page (on mobile)
• AI Window (separate from Normal and Private window, upcoming). Apparently it lets you chat with an AI agent to power-browse the internet.

No that’s way too paranoid. Honestly 20 years not 6 months. And by then ladybird will be viable so nbd

I wouldn’t mind a web browser being part of a broader system of trusted software, but shoving an AI chatbot into my web browser does not make me trust it more.

I like the accessibility features like offline page translation

The bubble is AI and they want some of that bubble investor money is my guess, so they put optional AI

And also … will the kill switch turn off the AI entirely … or partially? Since the AI system is baked in, will elements of it still operate in the background even if you turn off the switch?

In their defense a very tiny percentage of users even open options and of those an even smaller actually change stuff.

Maybe slighlty different for Firefox as probably more power user use it than other random programs. But basically if something is not enabled by default, it doesn’t exist.

All AI features will also be opt-in. I think there are some grey areas in what ‘opt-in’ means to different people (e.g. is a new toolbar button opt-in?), but the kill switch will absolutely remove all that stuff, and never show it in future. That’s unambiguous.

Sounds like they will be opt in, not opt out

Because they’re counting on people who know nothing about technology using the AI stuff when it’s placed in front of them.