Not a Spring Onion

@makram @Blender "It's boot-licking good."

@soatok CaaS: CEO as a Service "Using our hybrid team of AI agents and low-paid interns, we model all aspects of a CEO, like the decision making process or the erractic LinkedIn rants, to deliver the true CEO experience. Customize your CEO by buying them stuff like a house, a car, expensive watches. You can select from a wide range of skills, characteristics and addictions."

@Urban_Hermit @erik @MaddieM4 @Viss I especially love this "explain why you did that". Which the LLM obviously can't since it didn't extrude text on purpose. So after the LLM cosplayed an engineer or analyst or therapist, it now cosplays someone explaining some previous thought, by extruding more text that is plausible given the previous stuff.

@cwebber while(not_ported_correctly) { manually_add_guidelines(); }

@bri7 @futurebird [Not arguging that these models are 'thinking', even if it might sound like that.] I think the "explain how you arrived at that conclusion" that was all the rage is very interesting for two reasons: The modell is generating more text. It's not like it is showing you a walk through its model and the random numbers it pulled. So it is basically generating an explanation that is plausible given what the said before. I think this is often also a behavior with humans. My opinion about a topic might be a gut feeling, but when questioned I start thinking about it, trying to find arguments. Often ones I didn't already have when I stated my opinion. The first thing could make sense to ask if amodel are trained to change their position given new information. So they could "correct" a bad roll of the dice. Of course, a user might think that the model "really thought about this", which is obviously not the case.

@Cdespinosa Reminds me of the "NEO Humanoid". As if putting a human slave in a full-body-suit so owners don't have to look at them or treat them as human. "But it's just a robot." For which the company has admitted that they haven't the software for autonomous behavior (yet). So there is a person on the other end, manually controlling that robot. You can get one of these robots for $500 per month. Seems rather obvious how much that person will be paid and how their work will be valued.

@DocAtCDI I think they don't assume that you work on it full-time.

@ShadowJonathan@tech.lgbt "That's Universal Web Units, something like 'pixel'."

@macacator @PopTarts @alterelefant @Flowermob @Huntn00 @kibcol1049 "So, how did you become a hacker?" "Just trying to live at home."

@unnick@booping.synth.download Windows blushing in source code: >\\\\\\\\\\<

@alice@lgbtqia.space So Kiosk-kun doesn't get a tip? :neobot_bottom:

@silvermoon82@wandering.shop @ShadowJonathan@tech.lgbt @arch@floofy.tech "Tried to throw my developers a bone... but they immediately started fighting for it."

@kibcol1049 @Uair As I wrote: I had a rather privileged upbringing.

@Uair @kibcol1049 In my day, we mixed cocaine with chalk so it would be less suspicious. Still very suspicious when kids start snorting chalk powder in class.

@jschauma@mstdn.social But it's so much easier with AI...

@futurebird@sauropods.win @ireneista@adhd.irenes.space And that was basically the idea. At least that's how I understand the paper from 1950: Take something vague that can "best be answered in a poll" and make it testable.

@ShadowJonathan@tech.lgbt Tired: Did you ever stop thinking and forgot to start again? Wired: Did you ever start thinking and forgot to stop again?

@Uair @kibcol1049 After school, we had to work in the local cocaine mines for a typical 16-hour-shift.

@WooShell@chaosfurs.social @FoxesInLove@social.neonpaws.digital "Vore or polycule, you decide."

@nova@lunar.place Nova mocking humans by making gowronface again.

@ireneista@adhd.irenes.space @futurebird@sauropods.win Popular perception... "Einstein? Isn't that the guy who invented the atom and then took the job as a search mascot for Salesforce?"

@nova "I decide where 'up' is."

@futurebird @ubi Food from processed algae that grows in the walls and gets its energy either through natural light or heat (which could be produced from excess power from the solar panels).

@zynaesthesie @DerKlimablog Deswegen brauchen wir ja mehr Industrie. Damit wir mehr Öl kaufen können. Das wir dann verwenden, um den 6-Stunden-Tag einzuführen. Der dann unsere Industrie befeuert. "Arbeitsstunden sind für die Wirtschaft so wie Strom für das KI-Rechenzentrum: Je mehr, desto besser." (Kein Zitat, könnte aber eins sein)

@simon_brooke @screwlisp @vnikolov If Access Control Lists were so good, why didn't they make Access Control Lists 2? Oh, wait.

@mmiasma@mastodon.sdf.org Thankfully, AI doesn't have access to the airgapped backup.

@cwebber@social.coop I remember reading in a novel 20 years ago that two fictional worms merged in the wild, creating a super worm. "That's super unlikely from a technical perspective.", I thought back then.

@fogti@chaos.social @cwebber@social.coop Two LLMs talking to each other until one gives up (or enters a reasoning loop).

@DeliaChristina@sfba.social @grumpasaurus@infosec.exchange Internal note: "Replaced the user's sugar-filled keyboard with another sugar-filled keyboard. Enjoy."

@jonny@neuromatch.social Future ads will look like little noise PNGs, densly encoding information designed to confuse "agents".

@niconiconi@mk.absturztau.be "We would need a Terabyte GPU memory, because otherwise the models wouldn't fit."

@futurebird@sauropods.win pushes open cola can from desk That was close. Soda is bad for you.

@sjolsen@tech.lgbt Like a LAN power outlet? We recently bought one and as far as I can tell, there are only two categories: Expensive andIndustrial expensive wifi won the cheap segment because it is easier to set up and use. Sadly.

@kibcol1049@mstdn.social That should be it's own thing. "I am partial to doing projects..."

@spork@furry.engineer At least we know now who let the dogs out.

@yacc143 @dotjayne A few years ago, people could have argued that one needs to be especially careless and/or malicious to affect large-scale damage on a production system. In a way, I think it's very interesting that a entirely new danger category has emerged that is labelled "our own tooling". And since the AI tools are pushed by the companies themselves, it should in principle be harder to blame production problems on single employees. (In practice: "Why couldn't you magically forsee all problems?" and "Why didn't you check every line of that gigantic script?")

@yassie_j Mobile phones don't provide much mental stimulation anyway. As long as there is no rule that forbids playing an MMO while driving, I am fine.

@kibcol1049@mstdn.social @tsvenson@mastodon.online "The signup page...? Uh... we're still working on that..."

@4censord@unfug.social "Of course Dev 1.0 can't solve the harder problems. But we are training it to become Dev 5.0 one day."

@niconiconi@mk.absturztau.be If you are adjacent to too many bullets (pixels in the state "alive") your character might change from "alive" to "dead".

@gwenthefops@transfem.social Time to fry some french then!

@yacc143 @dotjayne Yes, I agree. My point was more: You could always blame a single person or human error. "This was not a systematic problem, that was this one person doing something very wrong." Not having an offline copy of your data is obviously a "systematic" problem at a certain scale. As is not revoking old keys, having a proper overview regarding permissions etc. I think that so little goes wrong because people are normally rather careful. And our systems are designed for careful people, not careless ones. Now we have tools that are essentially random, but it feels like "random with intent": Deleting one database is an accident, deleting all of them is having an accident, then following through.

@rrustema020 @notjustbikes pwned

@mattblaze folds out a huge paper with random stuff written all over it, amongst other things apparently featuring both 'cold fusion' and 'the pope' No, you see, you don't understand...

@lina@ohai.social But adam named all the things afaik...

@dckim@mastodon.social @inthehands@hachyderm.io In case of LLMs, neither the generated code nor the generated tests are sufficiently random, so... chances that another set of generated tests also "shows" that the code is correct should be pretty good.

@parents4future@climatejustice.global Irgendwann hat die Menschheit Konzepte wie "freie Wahlen", "Streiks" oder "Demonstrationen" als Alternative zu "ungeliebte Leute hängen" entdeckt. Ich bin dafür, dass wir bei diesen Errungenschaften bleiben.

@HonkHase@chaos.social Die Anführungszeichen lesen sich so, als wäre die Idee etwas neues, dass Faschismus sich auch immer aktueller Technologie bedient hat. Goebbels würde heute Youtube nehmen, damals war es halt der "Volksempfänger".

@lina@vt.social "You stand accused of downloading and then redistributing the BritDonnaAguilera model, which is able to generate music extremely close to that of popular pop singers."