I’m all for freedom of speech, but dude…

yeah, I’ve seen this bs multiple times now. Also tired of github rewriting commits and changing hashes when they can just be fastforwrded onto the merging branch. So for a while now I almost exclusively use the command line to merge.

And you’d think by the third .md they’d start enforcing a proper file tree structure. Nope, let’s clutter the root dir with 20 uppercase md files instead, along with all dotfiles that a dozen different tools already dump there unconditionally.

hehe you have the word ass in your private key, what are the odds

I don’t know you, but it doesn’t sound like a high standards issue to me, sounds like a lack of process. I’ve been a thorough reviewer before AI, at least thorough in the ways that mattered, not nitpicking formatting. And I’ll tell you an automated reviewer today can catch more things before I have time to confirm the first item I find. There are still false positives, but it’s still more thorough than I have time to be. Bc of that it really helps including a round of automated review before looping in a human, regardless of who/what wrote the code.

The main kind of review problems AI still struggles with are the project direction ones: “does it make sense to implement this/like this?”, “should this be a new package instead?”, and things involving tacit knowledge that often goes undocumented “last time we did this, someone had to access prod on a Sunday” - so that’s what I focus my reviews on. And the other area is if you’re writing UI code, whether it’s a web app or a game, it’ll also struggle to determine what “feels” good to use, so it’ll need a human earlier in the loop.

This kind of specification applies to the agent behavior or the agent behavior in a project. You don’t write it on every conversation, you write/review it once for the project or agent and let the harness include it on every conversation. You can even ask the agent to scan the codebase and pull desirable patterns out of it for future sessions.

The entire point of using an agent is to not have to write everything yourself, so that when you write “implement feature X”, the feature gets implemented in a way that makes sense in that project, is reviewed, refactored, and tested by agents, and it’s good enough to bring in a human reviewer.

See, that’s the issue. People letting it “drive itself” get worse results. You should be the one holding the standards and guiding the model, otherwise you will be frustrated.

My honest opinion is that it’s bad because a lot of people using LLMs have no standards and push the first thing that seems to work. Be mad at who’s at the driving wheel, not the car.

You absolutely can generate crap with agents/LLMs, and like a humans writing, the first draft will probably be subpar or maybe complete garbage. Every new session is a clean slate, that’s why putting effort in the documents guiding it is so important.

I don’t think there was a lot of people working on the rewrite. Most PRs are from bots. Original estimates for a manual rewrite were a small team working for a year or so, which puts total costs over $1M. Even doubling the token cost estimates, it was still cheaper than doing it manually by a factor of 2x-3x

I’ve seen plenty of code in my life, from humans and AI.

For the past year or so, these agents can code just fine most of the time, as long as they are given enough context (or have the tools to get it). Regardless of how many downvotes I get here, they really are capable of generating decent code. I’m sorry you couldn’t make it work yet.

It was higher sure, but anthropic also has some of the most expensive models out there. That cost could be 5x-10x less just by going with cheaper model providers (if one were to pay the API costs, not the case for bun).

bun 1.4 was released 3 weeks ago, btw

unironically, a better prompt does yield better results - shocking, I know

well, that depends entirely on your prompt/process, it can be done

yes, I get that and it’s a bad default, but not RCE like the post says

That does not allow remote code execution. An attacker still needs a vector like an SSH session to be able to control kitty. And at that point, if they already have a shell session, they don’t need kitty to do damage.

edit: giving access to other programs is not remote execution, but obviously not a good default.

yeah, unfortunately regulations have a bad habit to catch up with reality after the fact. By the time we actually have people in charge making better choices it’ll probably be late.

30 rockets a day? wtf

why can’t he just build bunkers, like reasonable billionaires

systemctl list-unit-files --state=enabled to check everything enabled that you might want to disable

as for running as root, AFAIK all systemd services run as root, unless they’re defined at the user-level; but IME setting services at the user level usually means a more involved setup to make sure PATH and env vars are correct.

is backrooms a secret NSA project after the success of backdoors?

They don’t need to teach “windows” or “excel”. Working with an “operating system” and a “spreadsheet” is the real value. One can quickly pick up any equivalent technogy as they need to. There’s no good reason to stick to the company that happens to have a monopoly at the time. In fact, it’s better for society that they don’t.

As for photoshop or any media editing program, I’ve never learned that in school anyway. That was just not a thing in any school around where I grew up. Maybe I missed out, but I’d say that elementary and high school curriculums are already pretty busy to focus too much on that kind of work. IME teenagers usually get into that kind of software at their own time and interest. I was into video editing and 3d modeling, but that was never in the curriculum, nor would i advocate for it to be.