no notes
Remote
Karl
@karl@infosec.exchange
mastodon 4.8.0-alpha.3+glitchEngineer. Used to work on industrial cybersecurity for money. Now helping to build *that* someone else's computer. Interests include Unix, security, cryptography. I know enough about sand to not let it do the thinking for me.
Non-technical interests are climbing, watchmaking, records, beer, and whisky. Oh, and MTG.
Will toot terrible puns in either English or French.
Trans rights are human rights. Fascists can rot in hell.
254 Followers
181 Following
50 Posts
Joined November 02, 2022
recursion():
In the weights:
No stable person surfaced. This result was weightless aka weak af.
$EMPLOYER finally sent me my personalized link to the "anonymous happiness at work" survey!
🤗🤗🤗
Thanks buddy, I'll pass.
Replying to
@sherbang@chaos.social @Hex@kolektiva.social @orange_lux@eldritch.cafe @saxnot@chaos.social @dibi58@this.mouse.rocks I work at a cloud provider and can confirm that running ai isn't as cheap as it sounds.
My take is that most models are so general purpose that they're very inefficient (versatile, yes, but inefficient). Think "sorting an array through a bogo sort" inefficient. LLMs trained for a specific purpose may be more cost-effective to run long term, but that's not the norm.
lame
Replying to
@fuzztech@infosec.exchange wait until someone writes a skill file to author raw commits directly into the .git/
Replying to
@iris_meredith@mastodon.social This post matches my anecdotical observations. The only way to wield LLMs with great results is through a thorough understanding of the system you're building.
Essentially it boils down to Kernighan's Law, doesn't it? Don't try to get an LLM to write something you could not grasp yourself, it won't be able to fix its shortcomings and nor will you.
That being said, deep understanding of systems (at least for me) takes time. Time saved by LLMs is rarely invested in more expertise, it's rushed into the "next thing" that needs to be built; and this is how the foundations of the knowledge of the system become shallower.
Very happy to see how much the gear train spins backwards. Until the indirect seconds hand complication is added, that is.
Damn this J80 movement is cool. I really hope I end up fixing it. Keyless works is next (tomorrow I hope), and then the balance (scary!). Fingers crossed!
Replying to
@AlbertBorneo@framapiaf.org Hey, comment ça se profile pour vous ?
Punaise quel enfer. D'abord moi, puis toi…
Replying to
@misty@digipres.club I honestly have never cared for the cloud. But I've been recently hired by a French cloud provider and this definitely is a factor in making be believe my work there is important and matters.
Replying to
@da_667 I have always been in favor of responsible disclosure, not because it works but because it's the *right* thing to do.
However, in the current day and age of LLMs, should I find evidence of vibe-coded shit in whatever the next broken thing is, I wouldn't mind FD.
Educating myself on aerial firefighting for absolutely no particular reason.
There may be a significant fire close to my home.
Replying to
@hacks4pancakes@infosec.exchange I mean, yeah but it doesn't matter. We're not the target audience, the idiots who got him through the ballots are, and they *will* fall for it.
Replying to
@mijndert @thomasfuchs And this is why I bought a roomba that isn't connected. It doesn't care about servers shutting down because it doesn't know about the internet.
Old tech is sadly sometimes more reliable.
Replying to
@nixCraft@mastodon.social How desensitized have we become to even discuss a flip phone sold half a grand?
Replying to
@AeonCypher@lgbtqia.space we've been known to hire English speakers, but speaking French doesn't hurt
Sigh. My laptop's keyboard is starting to misbehave. Some keys, sometimes, don't register. Notably, the space key and the enter key are flaky. I'll try to clean it a bit, but I don't have high hopes.
Pro-tip: it doesn't matter if pride month is over. You are still allowed to tell bigots to fuck off.
Replying to
@whitequark@social.treehouse.systems I did not know that. Dabbling in watch repair, this is yet another excuse to get a good microscope.
RTO to boost productivity, they said. This is our open space.
Replying to
@cpages@piaille.fr Je ne comprends pas la question. En général, quand tu refuses la démission de quelqu'un c'est parce que tu proposes de travailler avec cette personne pour résoudre les problèmes qui font qu'elle s'en vont.
Une autre tactique c'est de refuser la lettre de démission en main propre pour forcer l'employé à l'envoyer par lettre avec AR, ce qui retarde un peu son départ légal.
Replying to
@cpages@piaille.fr Ah oui c'est pas pareil si tu essuies un refus de Jupiter tu t'exposes à son ire.
Replying to
@AeonCypher@lgbtqia.space If you speak French, perhaps you can shoot a spontaneous application to Scaleway. There's an AI department here and your pedigree may be relevant for them to consider your resume, It's not the worst © company to work for in these times.
Replying to
@lina If it does not make you happy, and it sounds like that is the case, focusing on other things that will is sensible. We're all here for a short time, might as well make it an enjoyable one.
I claim to have almost 25 years of XP.
Replying to
@fourmiune @AlbertBorneo@framapiaf.org (Oui, ça c'est totalement conspirationniste, si les réacteurs des avions étaient source d'incendie cela se saurait. Bonne journée.)
Replying to
@whitequark I agree, which is why some would have liked them to return the money they already had received.
I don't have strong opinions myself about this because I lack context (I don't know for instance if that has been considered, and perhaps technicalities would have made it a logistical nightmare), but I can understand their point.
Replying to
FUCK YOU GNOME DO NOT CHANGE DEFAULTS LIKE THAT.
Fuck you.
Replying to
@nini@oldbytes.space @dibi58@this.mouse.rocks I would like to see research supporting such an assertion.
I dislike LLMs very much, which I why I don't want to make unsubstantiated claims that can be debunked by LLM supporters, which would discredit any other concern, however valid, that I have with LLMs.
Wait basshunter has released a song this year? Huh?
The shit show never ends.
@arstechnica@mastodon.social https://mastodon.social/@arstechnica/116993887661140995
My dad had avast installed on his machine.
Given how many dark patterns this crap has, I don't know what distinguishes it from malware.
My company (EU-based cloud provider) is interested in sponsoring OSS that would need cloud credits for $PURPOSE.
I'm not in charge of this so no guarantees, but I can relay some ideas, if you have any (I already suggested curl, but Daniel declined as they already have this kind of resources provided to them).
Replying to
@nini@oldbytes.space @dibi58@this.mouse.rocks there's an in-between where you still train an LLM but it's more focused, but in essence, yes.
RE: https://infosec.exchange/@da_667/116850806990941776
How can a cybersecurity blog be taken seriously if they're not using poorly photoshopped catte memes?
Quoting
please stop using dogshit AI generated images for your cybersecurity blog posts and research. Thank you.
Open quoted postReplying to
@fourmiune Il n'y a pas de lien de causalité mesuré, ni d'ailleurs de corrélation constatée.
Si les départs de feu de forêt étaient à proximité des aéroports (qui, je le rappelle, stockent des quantités importantes de carburant), cela fait longtemps qu'on le saurait.
Ce qui peut être affirmé sans preuve peut être nié sans preuve. À vous de fournir des sources tangibles qui expliquent que les aéroports sont la cause des feux de forêt actuels.
Replying to
@fourmiune Il n'y a absolument aucune base scientifique derrière l'idée que les aéroports favorisent les incendies, c'est juste absurde. Ce sont des hypothèses renforcées par des arguments pseudoscientifiques.
@AlbertBorneo@framapiaf.org
Replying to
@whitequark Some would have preferred if they turned down the money period, but it's something. :)