classified

5 ms·

1.
by classified 21h ago
> Irregular describes “the agent itself becoming a threat actor”; Talk about responsibility laundering. The state of affairs is reaching unprecedented levels of absurdity.
2.
by classified 3d ago
> features spouses expect to just work Never underestimate the risks of spousal dissatisfaction.
3.
by classified 3d ago
Outfits like TSA, ICE, Flock attract all the wrong people.
4.
by classified 3d ago
After years of propaganda war and boundless "creative marketing", Misanthropic and ClosedAI have lost all credibility. With trillions in investments on the line, it's not too far fetched to assume that those warnings have bee
5.
by classified 4d ago
Your private personal data and interactions are never used to train our foundation models unless you explicitly choose to help improve them. That's practically the opposite of what your editorialized submission title says.
6.
by classified 4d ago
Teens will just learn not to use the tattle taxi. Or hide their toy guns before they do.
7.
by classified 4d ago
dEtermination. Also quite confusing.
8.
by classified 4d ago
Who would have thought that writing like a human would become a moat one day?
9.
by classified 4d ago
Let's hear you say that after it plunders your bank account and frames you for murder.
10.
by classified 5d ago
You don't? Shame on you.
11.
by classified 5d ago
Being uppity to your betters is one thing. Being inconsiderate towards everyone is something else.
12.
by classified 5d ago
Are you commenting on the wrong thread? This isn't about somebody leaving.
13.
by classified 6d ago
It's the usual propaganda. They want to insinuate how powerful their models are and achieve government regulation. Win-win.
14.
by classified 6d ago
The part you're excluding with "mostly" is still just the laws of physics. And it works according to those.
15.
by classified 6d ago
> my back arrow first slows me down That's exactly what would happen ins space.
16.
by classified 7d ago
> waste THEIR electricity Thank you for helping us heat the planet.
17.
by classified 7d ago
This is very instructive. Now I finally understand why so many developers claim they are much more productive with AI coding. Merely writing code without adversarial fights is just not load-bearing enough. How could we ever live without thi
18.
by classified 7d ago
Well, they have to crank up the token spend somehow!
19.
by classified 7d ago
> models will be so persuasive and skilled in social engineering That's an interesting theory. I'll believe it when I see it.
20.
by classified 7d ago
No wonder so much of our software is rubbish when we treat the compiler and the runtime environment with so much neglect. By disrespecting them, we alienate the entities we rely upon most. I foresee a huge surge in software quality once the
21.
by classified 7d ago
In the good old days, twitter was the replacement helpdesk for the monster corporations. If your issue went viral there, it would suddenly be magically fixed, when the day before it was completely impossible and out of the question. YC shou
22.
by classified 7d ago
Yes, but for humans it creates friction. Seeing a captcha makes me think twice and thrice if I really want to visit that site so badly that I'll endure the suckage. LLMs don't care, for them the friction doesn't exist.
23.
by classified 7d ago
And since LLMs are apparently good at circumventing the absence of an API, there's not much incentive to add them now. APIs are for humans. LLMs just break through all the captchas and anti-bot measures.
24.
by classified 7d ago
> not being a simple case of intellectual property theft No, it's an aggravated case, since it's the same way they got all of their training data in the first place.
25.
by classified 7d ago
> it doesn't seem to be the case Based on what? Your crystal ball?
26.
by classified 9d ago
> they still can’t count the R’s in strawberry! My resident Qwythos-9B counts 3 "r"s in "strawberry". So even locally hosted models are catching up.
27.
by classified 10d ago
The HN title ("Intellectual Fly Is Open") made me think it's about some startup starting to sell subscriptions. That title should be de-editorialized. Was it LLM-generated?
28.
by classified 10d ago
> I'm curious about why LLM writing fails. Apart from the tasteless manipulations of the providers, it's mostly training data. LLMs output the average of their training data, and the overwhelming majority of humans are bad writ
29.
by classified 10d ago
At least this is more informative than "Thanks Andrew".
30.
by classified 12d ago
It's poetic that this photo is from a storage room for disused things. The font salad reminds me of the 1980s, when novel desktop publishing apps first gave every computer user a large selection of fonts.

More ›