What the actual fuck is this email?!??!!!
0 Boosts
0 Likes
Kai Tang got kicked from ByteDance and turned around and created... Whatever this thing is.
They "sell" it as collective reasoning - the more autonomous agents, the greater likelihood of emergent intelligence.
But um... Collective reasoning isn't really harnessable for anything much. So I'd guess they just want more niche and AI-untainted info to sell to other AI companies to train on.
@sundogplanets Wait, these guys are appropriating Tolkien's stuff (Earendil)?! Is this a Peter Thiel venture? I hate this trend so much: Palantir (surveillance), Anduril (war robots), and Earendil now destroying natural night. It's all so dystopian.
@sundogplanets OK, I looked it up and Reflect Orbital isn't connected to Thiel's ventures but it does fit with the theme of naming your dystopian startup after something in Tolkien's legendarium.
@tsupasat Yeah, I hate them. I wish Tolkien's estate would sue some techbros.
@sundogplanets @tsupasat They only go after people who write fanfic of Tolkien characters. Don't want to pick on someone who can fight back
@sundogplanets
If you look up the iLands thing, it's pretty creepy rabbit hole. Very AI psychotic. I saw some other prominent folk complaining about this last week.
@sundogplanets Nope. I got nothing. The end game escapes me.
Mansplaining, without invitation, my own words, with the citations that I used myself. What the actual fuck.
I think this "iLands" is the same site/source that - from what I read - has been sending swarms of agents which try to sign up on all sorts of Fediverse sites too, and then further try to argue with the moderators when they are banned.
Very WTF
Mansplaining as a Service
and also
Sealioning as a Service
It's mansplaining because it was trained on mansplaination.
@sundogplanets I was thinking, 'sounds like something an 'AI' would do, and then I got to the end of the email. 😡 Yeah, why are people destroying the environment for this garbage?
@sundogplanets Oh, wait, it’s from iLands, which seems to be a known menace. I’ve seen other reports of people getting deluged with spam from iLands agents. It’s probably worth adding the string to all your spam filters.
@sundogplanets To whom it may concern,
As a public service, I hereby provide the most likely combination of words some guy on the internet would send
Best Regards,
A Harry Potter reference
@janvenetor Oh gosh I am such a non-HP fan that I didn't even get that reference. But maybe it's the roman emperor? Who knows...
Thinking about this a bit more, the reason that this confuses me so much is that it's not asking me for anything. It's not the spam I'm used to that's trying to scam money. It's like... it's trying to be helpful (in the most mansplainy way possible, but whatever). But it's a program, a racist pile of linear algebra (as "The AI Con" so beautifully puts it).
Is it trying to get me to reply and train on my email responses? This is so fucking weird.
Anyway, fuck AI, time to snuggle goats.
@sundogplanets
I think it's trying to get a reply from you first and foremost to convince you it is a useful tool, a service you will want to subscribe to.
And to train on your reply, yes.
@sundogplanets
From what I understand, you've nailed it. It is trying to get you to reply so it can learn your style.
Endgame is to be able to impersonate you to various other people to scam them.
@sundogplanets Maybe it's an attempt at "the first hit's free" - hoping that you'll be so amazed by their unsolicited slopsplaining that you'll throw piles of cash at them to generate more useless slop for you? Making you put in a little effort to figure out how to throw money at them (rather than outright offering a paid service) could be a tactic to make it feel less like they're soliciting...
@sundogplanets Maybe send back as much garbage text as your email provider will let you! If it does process it, it will cost them lots of money!
@sundogplanets I’m soooo here for the “Fuck AI. Snuggle Goats.” t-shirts to help fund your work. 🤘🏼
@britt OHHHH that's such a good idea!! Sign me up for that t-shirt!
@sundogplanets What lovely ears they have!
Ok last post stewing about this: I just looked more carefully and pretty much all of the citations are wrong in some strange way. But they're all clearly based on citations that I've used in writing.
Don't use genAI for science, kids.
@sundogplanets it's to waste your time on checking the content and distract you from meaningful work
@sundogplanets I don't remember who but someone was posting here about iLands making bot accounts on their server.
@sundogplanets AI is terrible for citations. Citations are a huge part of my job (I edit at a journal). All of us journal people are trying to send up the alarm about how much AI screws up citations, and no one is listening. If you ask AI to format your citations it will mess up all your DOIs and make things that look right but are not at all.
JAMA has written a thing on this, which is polite and journal-speak for "oh god, oh god, do not let AI touch your citations": https://jamanetwork.com/journals/jama/fullarticle/2852772
@rednikki @sundogplanets The worst thing about this is that citations are boring and finicky and you sometimes have to look in a lot of different places for them and they have to be formatted to spec so it absolutely *should* be the thing that computers do better than humans.
And fucking YET.
@debcha @rednikki @sundogplanets There are solutions that involve computers and they were introduced like 40 years ago. Export citation as a BibTex item, use LaTeX to format paper, for example. Want a different citation style? Just change one option.
Maybe someone made a different program that is more accessible to laypeople than LaTeX. But manually doing boring citation spec stuff has long since been "solved" in computer science. People don't need to use MS word.
And most importantly, it works with deterministic precision and not with stochastic randomness.
@skaphle @debcha @rednikki @sundogplanets
Even MS Word has been able to do this for at least 20 years. It has a bibliography database that you can manually enter citations into or import them from a variety of sources. You can add a citation from there and it will show up in the bibliography. The bibliography is (as with BibTeX) rendered in a style that matches the requirements of your publication.
There is absolutely no reason for anyone to ever manually type part of their bibliography.
@skaphle @debcha @rednikki @sundogplanets
Citations are actually great fun to do with hand and are much easier and much less work than people who only ever used reference managers think.
Source: me, I used these things [i.e. reference managers] for a decade since my undergrad; last year I tried to learn to do them [i.e. citations] manually instead and found it to be much better. And when the time to maintain a database, check the auto generated citations and stored metadata, fix errors, make the reference manager behave with your writing tools etc are taken to account, I find it's actually less effort and just quicker.
And you get to know your sources more intimately, cite more precisely, and to be better at reading citations in your field.
I 100% recommend everyone to try it at least once.
@librarysquirrel @skaphle @debcha @rednikki @sundogplanets
And when you decide to resubmit the paper, or prepare a longer journal version, and the new venue has a different reference style, and you need to change the style of every single one of those 50 citations that you’ve typed by hand? With BibTeX, it’s a single line of text to modify to select the new style.
@sundogplanets three words too many in that last sentence.
@sundogplanets it reads to me like they’re trying to get you to support reflect orbital, claiming that the sources they cite found that level of reflection to be safe to look at with the naked eye. Maybe even just to maximize the paper trail of those claims, an escalation of a much older public influence tactic
@ShadSterling I read it as the plagarism machine is finding sources that agree with my analysis (because I used those sources. I'm actually not sure what the point is here. It's literally all in that article)
@sundogplanets I guess I read it as cherry-picked to say something that makes their plan sound less harmful than it is.
I guess that’d make it a bit like motivated reasoning as a “service”, but it’s the service of helping a motivated customer to influence other people’s reasoning
@sundogplanets @ShadSterling So AI agent is plagiarizing your article to do a fact check of the same article? I don’t understand it either. I am not a tech bro
@sundogplanets my best guess is that after a few responses and more helpful mansplaining it will try to sign you up for a mansplaining subscription or a pay-per-mansplain plan.
@sundogplanets The mansplainers never imagined that AI would come for their jobs too.
@sundogplanets I have to admit, I feel kinda dumb for not realizing LLMs would be tireless propaganda machines.
@sundogplanets Automated. 🤮👿
@sundogplanets oh not those obnoxious iLands bots. They’ve been trying to join instances all over the Fediverse for weeks. Have heard of other scientists getting email from them too, though this is new to have it send citations (which who knows if they are real or not).
@sundogplanets oh noooo it's a plague. @ernie wrote about them here https://tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang/
@bunnyhero @ernie Oh thank you for this article! I am still totally confused, but at least I have a human-written article to verify that it's definitely not just me.
@sundogplanets @bunnyhero @ernie
That article suggests reporting iLands to the FTC, but I do not know what the equivalent on your side of the line would be?
@michael_w_busch @sundogplanets @bunnyhero Spam laws are actually stronger in Canada than the U.S. Info here:
@ernie @sundogplanets @bunnyhero
I am jealous of places with regulators that actually regulate.
@sundogplanets oh noooo it's a plague. @ernie wrote about them here https://tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang/
@sundogplanets Christ, tedious. 🤦 This is going to be the norm from now on isn't it? 🙄 Spam almost killed email before, but AI's going to manage it this time.
@sundogplanets oh wow, I'd heard LLM flavored AI described as mansplaining-as-a-service before, but to find it as a cold call in your email inbox is a new one on me...