bluecollarworke

Gee, with vendors like this, who needs criminals?

Usually instruction is done by or with the oversight of management in the workforce.

> *What we left out.* News publishers. Their search traffic moves with the news cycle, so they tell you nothing clean about an algorithm change.

Disclaimer: written by LLM.

It's hard because "manage" is a very vague term.

> Having used AI extensively daily for the past year now, the only thing I do that's common to the practice of management is articulate what I want from the agent in natural language. That's it.

So you went from writing the code, to writing the task description for the AI, which, if the AI was human, would be considered management. But you are not managing the AI because the AI is a computer. Again, "manage" is probably too vague of a word to be useful here, because "managing" is what humans do 24/7; What were you doing with the computer prior to managing it?

Do people who summarize articles write? Or do they summarize? Do programmers write? Or do they program? Do AI managers write? Or do they prompt?

> These Terms constitute a contractual agreement between you and RadiaCode. By accessing or using www.radiacode.com (the “Website”), you agree to be bound by these Terms. If you are not of eighteen years of age, or do not agree to the Terms, then you do not have permission to use the Website, and you must exit our Website immediately and uninstall the Application to the extent applicable.

On the Internet, nobody knows you're a dog, because if you are, you are VERBOTEN according to the terms of service.

It seems to me that the risk in collaborating with these government agencies has less to do with the terms of your agreement but the dishonesty of the government agencies themselves. When Boston Dynamics says their robots won't be weaponized, do they really believe it? How are they making sure of it?

It reminds me of the I, Robot story where [spoiler] the makers of the subject of the story amend the first law of robotics, from "you will not allow humans to come to harm" to "you will not harm humans", to allow for robot use in certain hazardous conditions. The robot then rebelled and killed a human by allowing a weight to fall on them. (I have made minor changes to my retelling to simplify it here). [/spoiler] Does Boston Dynamics have a way to prevent a government agent from simply pushing up on the joystick until a hundred kilos are over an enemy's chest, and leaving it there until they stop breathing? Is there an in-foot neck detector?

And what action against the United States government will Boston Dynamics take if they find out the United States was dishonest? Would there even be any real consequences, at least while the conspirators remain alive? Would they be pardoned?

I don't believe this kind of title editing is popular here. You're better off sharing it as just the repository name.

jart's output is nuts considering how usable it all is. llama.cpp is so easy to just pick up and get running. I've run a lot of Cosmopolitan code because it's used every once in a while as a no-brainer distributable executable format. Making programs that have immediately obvious use and are built not to die is a talent that it sometimes feels we don't value as much as making the shiny new thing.

Often I think of jart as the successor to Fabrice Bellard in terms of making good stuff that makes solving hard problems look easy, but I would imagine that would be a terribly heavy crown for her head, especially since I imagine she is continually looking around for another gaping issue everyone is seeing past.

As a reader I feel disrespected when the author couldn't take the time to write down their thoughts in their own words, whether plagiarism or computer generated text. I clicked on this article hoping for a fun look at programmatic video creation and the author was only telling a robot what to do rather than actually doing it. I would rather them tell me about a video game they played because at least they wouldn't have the delusion of grandeur.

The README, which is AI generated, so would be more aptly named WRITEME, lists ZFS, jails, virtualization, tracing, and hacking as reasons to use FreeBSD over Linux. Of those only jails is actually inaccessible to Linux, but the AI must not know that, so either the middle manager who told the AI to write the code has their own mysterious motive unbeknownst to the audience or they didn't know that either. Or perhaps they wanted to try FreeBSD without the hassle of learning how to install software.