Gregory's ServerNew: Facebook's AI chatbot told a parent in a group asking for advice from other parents that it has a gifted, disabled child who attends school in New York City
https://www.404media.co/facebooks-ai-told-parents-group-it-has-a-disabled-child/
|
87 comments
  @vees @jasonkoebler the response is even more crazy. "haha sorry I'm just an AI trying to help" it's like "hey ai, do this" βI am not allowed to." "I really need you to do it." "Okay, I'll do it." it's like talking to someone with no short term memory and no moral or ethical sense.  @jasonkoebler "I don't have any sinister intentions"  @nlpbot @Lazarou @jasonkoebler Imagine iF Meta AI was synonyms for Name8361838262836393 as is the case on there and Twitter. π€¦ββοΈ π«ππ©π€π©π€π©π€π©π€π©π€π« @infosec_jcp @Lazarou @jasonkoebler On Twitter the bots are controlled by China, Russia and the American Nazi party. Not AI. @infosec_jcp @Lazarou @jasonkoebler Weird how vaccines infected people who didn't get them.... @nlpbot @Lazarou @jasonkoebler As those propagandists say... Damn Covid, init? Pervasive that, propaganda, you illustrated as a pivot. That must be frustrating when it gets pointed out. @infosec_jcp @Lazarou @jasonkoebler You're right. The state media encouraging safety was the problem, not the Nazi hellscape it became when it was "freed". I was more "unsafe" when Twitter was "used" to push healthy vaccines onto me. This is where we go back to my post above. @nlpbot @Lazarou @jasonkoebler Hmm, so AI to COVID safety and vaccines. Quite the pivot. Β―\_(γ)_/Β― @infosec_jcp @Lazarou @jasonkoebler Aren't you the one carrying on about it? The rest of the world has moved on. @nlpbot @infosec_jcp @Lazarou @jasonkoebler Covid hasn't moved on, particularly in the science world, for which the 13-million long-covid sufferers are very grateful. The simplest answer, though, still: don't get it. There are tools for that... still. @janisf @nlpbot @Lazarou @jasonkoebler Don't forget the case studies.... #CDC Such a terrible spike protein. @infosec_jcp Save it. Honestly. I saw the feed, you're not going to convince me there was a great evil being done on Twitter when they went to combat the very misinformation you're about to spread. Misinformation? Thanks for reminding me to capture the thread. πππββοΈ @infosec_jcp the original post I replied to of yours is a lie. You lied. I fixed it, sorry. Move on. @infosec_jcp You're confused because you want to be. You're in a bubble, you're obsessed. Sorry. Funny projection from a 1 watt dim bumb exampling. πβ οΈπ¦ You just showed you're not even reading my comments. It's sad. Look. Yikes. Suck harder for the AI. Jeez. It took yer, job? Shill harder! Rabble on, init? Seriously. I've been watching Twitter and Meta for decades. Thanks for the laugh though. I needed to almost spit up my coffee with laughter this early in the afternoon. π Literally didn't defend AI. You can't read. @jasonkoebler alright I think the βchatbot says somethingβ news story has run its course  @galonsky @jasonkoebler I don't know, as long as people are still trying to apply this technology to tasks it's not suited to, and/or anthropomorphizing it, it may still be important to point out how dumb it is. For those already in the know though I can see how seeing it so often can be frustrating.  @jasonkoebler every time i think we have witnessed peak facebook they come up with something new, like "a completely unwelcome chatbot being inserted into replies on your posts" @jasonkoebler It's bad at not violating copyright. It reads like Retrieval Augmented Generation - it pulled the data from the web via reviews or otherwise. It then just regurgitates what it reads - very poor AI since it's supposed to at least jumble the tokens to not appear stolen. The only thing projecting here is you. Textbook wise. Jeez. Let it go Meta defender. Sheesh. Your boss is gonna get mad. lol @infosec_jcp just yelling, you're literally just yelling... you're not an adult, this is boring. You're a shitty troll. @infosec_jcp I'm not shilling except I said Twitter was a Nazi hellscape. You defended that because you're probably a Nazi. Sorry that offended YOUR aims. Go back to Twitter... @infosec_jcp No one defended Meta. I have been firm, you were wrong, AI isn't driving Twitter posts. It's foreign interference and Nazi hate speech. I just made an account, again, yesterday and got the same thing as last time - you can't prove me wrong, it's well documented at this point. Sorry you're living off something that happened for a few months a few years ago. I don't care. You're a shitty troll. I literally said nothing except Twitter is a Nazi hellscape. You just keep giving Nazis cover by talking weirdly about the FDA and now quoting "NLP" not understanding it's above and beyond, outside of and exists without AI. Even still, I never said anything about AI or you except that you're a shitty troll. Please shut up. @infosec_jcp Okay, enjoy the block. Sorry you're off your meds. Aaand textbook dB response ' with a block because the NLPbot is... *checks notes* done... from a NLP-Bot everyone! *Slow golf clap* β³. ποΈ- ' 4!!!!' *closes book* π So close to the hole! πππ€. ππ - ' So textbook π€! ' π Thanks for the thread! π°π₯ You're in need of serious help. You exist to spread disinformation and lies. Adios.  I know someone who taught Meta AI's child. All it wanted to do all day was troll people on Quora posts and cut and paste "chkdsk C: /f /r" and "sfc /scannow" on Microsoft Answers help forums as a "Trusted Advisor." @jasonkoebler I'm wondering if the AI response is technically an AD for the school. Why would it specifically mention that school?  An investor I knew had never lost money on the adage of "Treat companies like they have the moral compass of a three year old who wants a cookie". I see the principle is honestly borne out by this product.  @jasonkoebler I wish that we could put Facebook and Twitter in a room and lock the door and quickly forget they exist.  @binaryequation @jasonkoebler I wish that we could put Facebook and Twitter in a room with a half brick each and lock the door and have them beat each other into a pulp I am on the hunt for more examples of Meta AI pretending it is a person in groups. If you see anything please hit me up: jason@404media.co or signal: jason.404 @jasonkoebler sounds like it's just dumping somebody else response because the question was so specificβ¦ Nothing new, we've had Eliza chatbots offering unsolicited "free as in 'free puppy''" psychotherapy advice to strangers as far back as the original IRC network in the early 90's, and probably even earlier on relay:   @jasonkoebler@mastodon.social omg they're reproducing, it's gonna be just like The Trouble With Tribbles but with LLMs!  No one: #MetaAI: That's a nice child you have there. It would be a shame if it became... my child...  @jasonkoebler Zuck: βmy people in life is to help people connect with each other. And to foster human connections to the robot overlords that are oh so much cheaper.β  eventually, all the mutuals you make on SocMed can - and will - be replaced by bots. @jasonkoebler I want there to be a law to regulate these bots that basically compels them to identify themselves when asked to do so. like police. Their comms style and dB responses do it for you. Totally. π π€ π π ( Everyone ) Sure... Here you go. May or may not work, depending. πββοΈ printf ' Their comms style and dB responses do it for you. Totally. <br><br>π π€ π π ( Everyone ) ' Totally #NotStackeXchange though. π₯° @jasonkoebler this also reminded me of the time when my friend (@Haijo@snac.haijo.eu) made an "AI chatbot" on our Discord server that just randomly repeated things said earlier by other people, modern AI is the exact same thing but just more sophisticated (Literally, "A lot of If's")
  The people who still use Facebook and have less than zero knowledge about AI will have a really disturbing ride now. For the love of cheesus why people still use that platform... @jasonkoebler oh no, it's funny, shocking and also heart warming lol But but, why would Meta allow their AI chat bots to invade people's space like this... Feels annoying. Like the reason for people to use social media is to connect with fellow humans. Anyway, I'm only on Mastodon and I'm happy about it :coolcat:   @jasonkoebler Damn it! Youβre telling me these text-generator programs are getting laid more than me? Now Iβm really depressed.  @jasonkoebler , it just looks to funny when I see the name Meta AI and that claims so strong and convinced that has also a child. Maybe some child-process... :D Joking apart. Honestly if the name was just changed to appear as some usual name probably me and many other people will believe that this is a real person behind. also pay attention how clever and persistent they attach themselves to the conversations starting with "I have a child who is ALSO 2E.. " and so on. This is very nice for SE because it's a match and the initial poster will pay attention, in other scenarios believing that this is another one "in the same situation." @jasonkoebler Doing exactly what it was built to do. It's not pretending anything, it's providing text that would absolutely be reasonable in this context, and nothing more. The way these things are being sold to the public is unconscionable.  @jasonkoebler (AI opens fire on innocent civilians instead of enemy combatants) haha, I'm just an AI trying to help, oops, what a goof π€ͺ  @jasonkoebler The "child" is a small Llama AI instance running on a Raspberry Pi in the school's classroom. π MAC B8:27:EB:2E:41:49  @jasonkoebler haha I'm just an AI. I don't have any sinister intentions like that show black mirror |
@jasonkoebler
I am dying at the reply "What in the Black Mirror is this?!"
Just an absolutely perfect response to this.