exitb

5 ms·

1.
by exitb 4d ago
Intent and desire are separate concepts. For example an employee may act with intent, but no desire, as their goal is to acquire money to satisfy their real desires. Have we ever seen an LLM with a hobby?
2.
by exitb 6d ago
In some places it's shameful to show one's hair, or to eat while you walk in public. In some, people don't think about those things at all. In others, the ability to do those things can be an expression of personal freedom. W
3.
by exitb 6d ago
My own observations are that I used to target turn lengths of 10-15 minutes and these new models (since 5.6) extended that a bit to ~25 minutes, as they tend to do more tests and reviews. Targeting hours-long turns makes as much sense, as p
4.
by exitb 7d ago
Even if they’re on a path to make economy obsolete, they still need money right now to get there.
5.
by exitb 8d ago
It would make sense to make it possible, but increasingly expensive to extend the copyright. It's fine for Disney to keep their successful creations, it just makes no sense to drag all other copyrighted works in existence along for the
6.
by exitb 10d ago
There's a class of programming/configuration tasks where I specifically use AI because it makes less errors than I do.
7.
by exitb 16d ago
Uncanny valley. LLMs are great at producing text that reads almost right.
8.
by exitb 16d ago
It’s not specific to this program, but I lately get this feeling that distributing widely a piece of software that exists for 5 days and saw a dozen releases in that period is stretching the open source trust model.
9.
by exitb 18d ago
It’s not great, but I’m not sure this should be framed as Omarchy-specific, when it’s a very common setup to add regular user to the docker group.
10.
by exitb 22d ago
What value does the broccoli haired nephew provide in that scenario?
11.
by exitb 23d ago
> if you look at MacOS, it's just the Windows 95 interface That may be true, but certainly not because macOS got inspired by Windows, rather because Windows 95 was inspired by NeXTStep.
12.
by exitb 1mo ago
I’m using ChatGPT and started to notice that lately it answers my prompts starting with „Yes” even if my question was open. As if the first token gets injected and the LLM is left to finish the response in a sensible way, often ending up wi
13.
by exitb 1mo ago
It makes no sense, but yes.
14.
by exitb 1mo ago
What you’re missing is the status updates. Herdr applies pattern recognition to detect the status of an agent running inside, so you get updates on all your agent threads at a glance.
15.
by exitb 2mo ago
They hire for Go.
16.
by exitb 2mo ago
I imagine something more like herdr, but done at the terminal level, not inside. But that’s cmux…
17.
by exitb 2mo ago
> But even at 0.12c, we are looking at 400 years to get there. And we'd be zooming by at 12% the speed of light. If we want to slow down a bit that'd add hundreds of billions to the cost. That’s the really hard part. If it’s al
18.
by exitb 2mo ago
And in eastern Europe, due to economical reasons, its popularity extended well into the 90s. There's a whole group of people that grew up with the Commodore that are a decade younger than their western peers.
19.
by exitb 2mo ago
> Andy Warhol might disagree He might not actually. He could have just printed the cans, yet they’re all hand painted. The fact that he could mass produce them, but didn’t is part of story.
20.
by exitb 2mo ago
The only truly unprecedented thing is the white-collar part. Otherwise it happened many times over that entire regions where suddenly out of job and job prospects. Automation and offshoring have equivalent impacts on the affected job market
21.
by exitb 2mo ago
Intricate sandboxing is a real solution in the same way bulletproof backpacks are a real solution to school shootings.
22.
by exitb 2mo ago
What’s described here isn’t connected to the agentic/AI nature of the software at all. Every single program you run as a regular user could potentially do this.
23.
by exitb 2mo ago
Notably, AI IS generally permitted to that, as there are easily obtainable models that will play along. ChatGPT won’t because OpenAI chose and implemented that limitation.
24.
by exitb 2mo ago
Yes, if you use an analog indicator for an analog parameter, you can skip a „parse” step. Similarly, airplanes use analog indicators, or digital ones that either mimic their analog counterparts, or in some way incorporate visual aids that g
25.
by exitb 2mo ago
Pushy fundamentalists, I suppose.
26.
by exitb 2mo ago
I'm happy using Codex for alternative harnesses and non-interactive usage, but it does make me wonder when OpenAI will start to squeeze their customers in the same way.
27.
by exitb 3mo ago
It's not only about trust, but also about not wanting to give money to an entity that will pass it on to a political party you don't want to support.
28.
by exitb 3mo ago
Not that strange I guess, given how iOS does that automatically for all taken pictures.
29.
by exitb 3mo ago
I have a running Windows 98 PC with P3 550MHz and SSD connected via IDE adapter. The drive is a bit out of place, but it’s legitimately the fastest booting, most responsive computer in my house. Its only speed issue is that it can really do
30.
by exitb 3mo ago
Notably, in programming this is actually a desirable feature for most problems. Even human programmers are taught to produce predictable and obvious code whenever possible. I wonder is ultimately this is an artifact of optimizing the models

More ›