ai haters think I'm a booster, ai boosters think I'm a hater
engineering and infosec · he/him (decaf) · vis tacita
maine bluesky feed: https://bsky.app/profile/did:plc:txfqncx66asrjzitxfur3of6/feed/aaap7ldsnvpkw
THE INTERNET IS SCARY: @internetisscary.org
Technology exists and magic does not. I don't mean that flippantly, there's a lot there if you start to unpack it. I think existing is kind of the only difference. People who manipulate technology are basically actual wizards.
My specialization of the wizard class used to deal with arcane runes like `(() -> {})` until the text elementals were discovered. They possessed the machines, and now we mostly whisper prayers to them for favor.
Technology exists and magic does not. I don't mean that flippantly, there's a lot there if you start to unpack it. I think existing is kind of the only difference. People who manipulate technology are basically actual wizards.
If you have fairy sparkles that can make flowers grow, even if they are of the natural world, it is technology to learn how to wield them. Fire is a natural phenomenon, summoning fire on command and directing it toward productive purposes is technology.
yeah. my general thesis is that a lot of people have never had or given themselves the opportunity to be bad at things that other people can see and evaluate
One of the things about working in security is that sometimes you have to just sit on a really juicy rumor that could get you a lot of clout. But saying "There is a vulnerability in the Hello World app" discloses to the attacker that they should search for a vulnerability in the hello world app.
You can point an agent at a tweet and say "figure out what this is talking about" and it usually will do just that.
If you want the clout, post the checksum with no clues and then promote it later. That's the behavior of a professional.
Every US State’s Favorite State Besides Their Own (With Arrows)
Credit: Matt Shirley
It's surprising how few states reciprocally like each other.
Full list: brilliantmaps.com/fav-us-st...
This is absolutely true and totally infuriating. It's the most Star Trek technology to come out in decades and it's getting backlash because of all the narratives they could run with, they picked fucking Competitive Kyle from the Charts and Graphs department.
like love it or hate it, one thing you would be disingenuous not to acknowledge about ai is just how aggressively supply-pushed it is and how utterly tone-deaf the marketing messaging tends to be
They could literally just show clips of officers on the Enterprise going "Computer, survey historical occurrences of yadda yadda" and the computer going "Processing..." -- and like, we can do that NOW!
The first genuinely honest ad for generative AI I've ever seen, from Microsoft / CoPilot
A mediocre white guy sees his non-white colleague getting praised for good work, he asks AI to *literally copy it*, and then he gets all the praise instead:
www.ispot.tv/ad/SoDi/micr...
I think about this a lot, mostly in that I see it as an opportunity. There has not been a better time in living memory to apprentice for someone who's trying to retire.
The general greying of society will lead to a lot of things happening simply due to less young people that various people will claim are clearly actually due to whatever bullshit they’re on about. It’s gonna be very irritating.
Small colleges will collapse which will be the based youth rejecting Marxist indoctrination or the shameful abandoning of education by late capitalism, when the reality is that there will be less 18-22 year olds.
Military recruitment will become more difficult and this will be spun as everything from the youth rejecting the imperialist war machine to wokeness choking the military when there will simply be less potential junior enlisted in the general population.
The general greying of society will lead to a lot of things happening simply due to less young people that various people will claim are clearly actually due to whatever bullshit they’re on about. It’s gonna be very irritating.
If you ask a question of a machine, and you get the wrong answer, then you asked the wrong question. Blaming the machine for being faulty is akin to shouting at a rock for being in the way of your toe.
The argument I'm making is almost tautological, to be fair. If you give an agent direct access to query a database, and in the course of its queries the agent deletes your data, then you gave bad inputs. The database access tools you choose to give it are part of the I/O loop.
This might just be a trap for my blocklist but I'm really curious what people who are heavily using AI in software jobs and consider themselves comfy with it think AI literacy looks like. How do you encourage it for juniors? What patterns or habits of work are you trying to encourage and use?
I think this is one of the things that throws people about AI. We aren't used to a conversational counterparty that can't take accountability for misunderstandings. Your work, as an AI wrangler, is to make it difficult for the machine to misunderstand you.
Even if the instructions seem plainly and obviously correct. If the machine does the wrong thing, it is because the instructions were incorrect. That is the definition of "incorrect" when dealing with machines - instructions are incorrect if machine interpreting them does the wrong thing.
Because the machine, by definition, cannot misunderstand you. It can only do linear algebra. The consequences of that linear algebra depend on what you gave it for inputs and what you did with its outputs, and nothing more.
If you’re going to say that we need to “act on it” you should share some opinion about what that action should be. Saying “Someone should do something! [gestures vaguely]” is unhelpful.
One of my armchair theories of the last decade of politics is that America as a country has always had a central unifying belief that we should overthrow the monarch.
And unfortunately you can’t do that if you don’t occasionally elect someone monarch-shaped.
No joke. I think this is the most important video I have ever seen about mobile games. This guy breaks down how this game psychologically profiles you, and makes the game harder or easier in order to show you ads
A language model is no different from a stochastic parrot in the sense that they both produce the same externally observable behavior, but they are also quite different in the sense that a language model exists.
If you asked me to formally define the term "stochastic parrot" I would define it as:
An imaginary object that, if it existed, would do the same things as a language model, but by virtue of not existing, is harmless to any philosophy or sense of self.
Why would a physical search of Omar’s home in 2026 be relevant to or necessary for an investigation into whether her dad filled out a form incorrectly decades ago?
I don't think LLMs are conscious but I do think (and I don't know if there have been studies on this or if this has been verified, it is just something I think is likely) people who regularly use LLMs and also frequently behave extremely abusively to LLMs are more likely to be dicks to humans too
When I'm working with coding agents, I say things like "I'm sorry, I still don't understand..." and "Thanks for your patience" for because:
1. I think it models a more effective and professional conversation, which is the register I want to work in.
2. It's a good habit.
If you're feeling frustrated when working with machines, it's a good habit to recognize that the frustration is happening inside of you, irrespective of the machine. The machine is not frustrating you. You are frustrated. You are the adult in that scenario and part of the work is managing yourself.
Once again, if you are a senior researcher spreading doom to junior researchers, shut up shut up shut up. It’s your job to figure out a path forward, even in high uncertainty times. Come back when you have something constructive.
Huh, there ARE no science-fiction novels written about largely inert and harmless AI. Guess we don’t have real AI yet! Let’s consult our map of SF-plot-worthy AI and emulate that: