This is great news, and I might be tempted to use it if I had some reassurance that the mail servers (and the organisation that controls them) weren’t subject to U.S. jurisdiction.
Finally, Mozilla is finally doing something innovative for once. Stalwart is freaking awesome.
Thunderbird is going hard on Rust all of a sudden. Recently they started using it in Thunderbird and now they’re starting a whole service in Stalwart (written in Rust).
Super awesome.
Inb4 Mozilla dies because the lead of the AI development division is the ex-ceo who followed up massive layoffs with doubling her multi-milion salary every few years.
Collecting more data to sell them to 3rd parties?
If you’re referring to that fuckup with the ToS or whatever, that’s not what they’ve started doing. You can verify this by their Privacy Policy, which hasn’t changed in almost a year.
But if you are pointing to other examples, I’m open to learning.
Exactly to that. But I don’t care. Believe what you want. I’ve DeMozillad myself long ago. Couldn’t care less.
I mean, you clearly care
That’s fine. Do what you want. I’m not here to judge your choices, just point out that Mozilla only fucked up the communication, not the policy itself.
Next I hope Thunderbird’ll restore the GTK/Qt theming support they removed two years ago…
Ok, this part is pretty cool:
Thunderbird Assist will also be available. This experimental feature, developed in collaboration with Flower AI, offers optional artificial intelligence functionalities for users who want them while also addressing privacy concerns head-on. On devices robust enough to handle AI models locally, Thunderbird Assist processes everything on the user’s own machine.
However, for users on less powerful hardware, the development team has integrated NVIDIA’s confidential computing to keep any remote processing secure. Rest assured, those who prefer to skip AI services can continue using Thunderbird without these extras.
I’ve been unwilling to touch cloud based AI, much less expose my emails to it as there’s no guarantee of privacy, but being able to run a local model allows you the functionality without the risk. Haven’t used Thunderbird in years, but this is tempting me to give it another shot.
What’s that NVIDIA’s confidential computing?
Why on earth would I want AI integrated in my email?
I think I’d be pretty pleased with that actually, so long as it’s on my local machine. That’s because I often find myself wanting to locate a particular email that is along certain lines, or on a certain topic, or involves an organisation’s name that kinda sounds similar to this one word but isn’t actually that word or things like “the email where they mention they’ve had a kid” but I can’t actually recall either what they called their child, or what gender they were, or when the email was received. Or actually, even better, in that last example “What’s Dave’s kid’s name again?” and just getting a 1 word, correct response, with the ability to open the email it found where this was mentioned for additional context if I want it. Or things like “how long has it been since we moved out of that house?” and instead of finding the earliest email I can on the topic of moving house and reading emails to surmise when we discussed leaving and then finding which one might have mentioned that actual date we moved out, I could just get an answer, in English again hopefully with a link to the email or emails that provided the rationale for how the answer was arrived at.
Often in those simpler search situations I mentioned where I just need to find a specific email, keyword searches don’t always cut it. I have an absolutely appalling memory so figuring out pertinent details to things happening now based on what was going on in my inbox at some point in the past are a very important way that I get by. If I could achieve this more easily by asking relatively vague, English language questions that will help direct search efforts that are being done for me would be really helpful. Sure, theoretically all existing means of filtering and searching email should eventually find me that message but they’d likely be more effort than just asking directly like you’d ask a person tasked with digging through a filing cabinet for you, and sometimes even after extensive filtering by all kinds of clues: date, senders, keywords, labels, subject lines, emails I remember around the same time that I can find; I just for whatever reason can NOT dig up that email only to discover it later when it’s too late to be useful to me anymore and get to see what obscure reason it was none of my clever search methods caught it…
“AI” or as they were called for the longest time machine learning algorithms can do things like spell check and help with grammar.
The more modern algorithms that they started really calling AI can help format your ideas, can fix sentence structure, and can even translate into foreign languages
Email is probably the most useful place for AI as most of the ones we talk about today are really good at language formatting but don’t really have any intelligence
For example you can write an email cursing out your boss saying “as I fucking told you yesterday” and then ask the AI to rewrite your email in a professional tone so that it says “per my previous email” like sure you can obviously do that yourself but it’s a lot faster to word vomit your thoughts into a computer especially when it’s trivial work related garbage and save your mental energy for your personal time
Trying real hard not to be old man yelling at clouds, but have things gotten so bad people can no longer write a simple email without help with sentence structure?
I’m a software developer, not a writer or a salesperson, but I have to do sales to sell my software.
I can write a first draft of a sales email to get my ideas across and then have the AI look at it from a specific perspective I don’t have the skills in.
I dont just take whatever it says and hit send though, I have a conversation with it to tweak things i don’t like, remove things that I don’t think are needed or add things it missed.
Do this for 15 to 20 minutes and I end up with a much more polished email that won’t come across as AI slop with all the personal touches I did want to add.
Just a question on the value of time: If you can’t be bothered to write it, why should anyone bother to read it? Is it really that valuable of a message?
I do write something, and then work to refine it. Like I said, I spent 15 to 20 minutes on it after writing it.
Don’t tell that you need so much time to write an email 😳
We probably never sent as many emails as we do today, and there are bigger priorities. Wanna blame someone, blame the overly pompous corporate world.
Well I can’t wait for AI to write an email for a coworker that’s full of misinformation. I can’t wait to waste my time with that!
You still have to double check for mistakes. But even that takes less time than writing a 3-4 sentences semi-formal email from scratch.
In any case, your emails aren’t safe either, mate. Even if you don’t use AI, the person on the other side most likely will and your emails will be scanned just the same. Nobody is safe. Which is why this Thunderbird feature doesn’t excite me in the least.
Please signature the email with “Sincerely, <your name> and <name of the AI you used to write this email>”. Otherwise people get very suspicious when they meet you in persona and you come along differently than in your email.
You don’t, neither do I. why the hell are they pushing this?
It the jingl8ng of the keys that may sound good to some investors?
Do you want a serious answer or are you just being flippant?
Serious. I can’t think of what good slapping AI in there does. Is speech to text that’s existed for decades considered a LLM and therefore AI?
Other guy covered it decently. I think it’s also remember to have a little bit of humility and remember that just because you don’t have a use for something doesn’t mean others don’t either. In addition, just because you can’t think of a use for something doesn’t mean there isn’t any. I say this as a huge critic of AI/LLM’s and how they’ve been deployed.
No, I don’t think humility is the problem. It’s just a bullshit idea that nobody needs.
AI audio tools have allowed me to restore audio that five years ago I would’ve said had to be re-recorded or is otherwise unsalvageable.
AI transcription has been one of the greatest gifts to the disabilities community we have seen in decades. It is borderline as significant as screen readers for the visually impaired.
Again, just because you don’t know about applications doesn’t mean there aren’t any. You need to stop speaking out of turd about things you don’t understand. Again, I say this as somebody who is harshly critical of LLM‘s and the theft that trains these systems. There are a lot of different forms and tools that this technology takes.
You’re going to do all that in your emails?
Well, i’m interested in a serious answer?
Summary, auto-correct, translation, text-to-speech, speech-to-text
I must admit its been a while, but all these things where horrible last time i played with them.
It’s a *lot* better these days. Here’s how you could use it to improve the English in your comment:
Can you point out and characterise the grammar mistakes in the following sentence, with tips on how the writer could avoid similar mistakes in future - “I must admit its been a while, but all these things where horrible last time i played with them.”
Sure — that sentence has a few grammar issues, each pointing to a common kind of slip. Here’s the breakdown:
- “its been a while” → “it’s been a while”
Mistake: “its” is a possessive pronoun (like his, her, their), while “it’s” is the contraction of “it has”. Fix: Use “it’s” because the writer means “it has been a while”.
Tip: When in doubt, try expanding “it’s” to “it is” or “it has” and see if the sentence still makes sense. If it doesn’t, you probably want “its”.
- “where horrible” → “were horrible”
Mistake: “where” (a location) is confused with “were” (past tense of are). Fix: Use “were” to correctly form the past tense of “to be”.
Tip: Say it aloud—“they were horrible” sounds natural, while “they where horrible” doesn’t. If it’s a verb (they were), you want “were”.
- “i played” → “I played”
Mistake: Lowercase “i”. Fix: Always capitalize “I” when it’s a pronoun referring to oneself.
Tip: This is more about habit than logic—just train yourself to always type “I” in uppercase, even in casual writing.
Full corrected version: “I must admit it’s been a while, but all these things were horrible last time I played with them.”
And with the exception of the first, commonplace for a long time.
I would also argue the first use is actually a process flaw if implemented. That’s a great way to make sure no one reads emails, and all kinds of information is lost
They don’t list which stolen data Flower Ai was trained on. Ai slop is Ai slop and will only lead to less critical thinking from the general population: https://www.microsoft.com/en-us/research/wp-content/uploads/2025/01/lee_2025_ai_critical_thinking_survey.pdf
I keep saying this. As if humans aren’t dumb enough and lazy as hell without AI… I’ve met maybe 1 younger person out of hundreds that actually shows interest in learning and work ethic (the good kind, not boomerism). I’m not even old, but it’s looking like a dark road for humans. Now Fuck You, this is Carl’s Jr!
I think that a lot of the “laziness” comes from alienation from your labour. People are going to work the least amount to get their paycheck and if their job doesn’t account for the effort they put in, they aren’t going to do anything but the bare minimum.
Your points are valid, but I look at AI as an unavoidable trend in the tech space, which is why I experiment with local models. I’d rather understand how they work (and how to protect oneself from them) as I believe avoiding them isn’t really possible in my field.
Thus far, the local models I’ve worked with have gotten a C- on coding, but an A+ on bullshit. I think the tech still has a long way to go before it lives up to the hype, both negative and positive.
Thus far, the local models I’ve worked with have gotten a C- on coding, but an A+ on bullshit.
I work at a company that won’t allow us to use a search engine but has a local model we’re allowed to use, and this is a pretty apt summary.
How do you get anything done at all??
Right? I just do my best to ignore the bot and only enter queries any first year CS student would know. The rest comes from my memory and a few bookmarks I have saved.
s/chatgpt/journalists/too.
