I just actually thought that many of the great artists absolutely did paint, sculpt, or draw their creations "by hand", themselves. I suppose they used tools: the brush, chisel, pencil. Not sure how comparable those are to having an LLM write code for you.
The reality is that many if not most of the big names had apprentices who did a lot of the actual work. Art historians spend a lot of time arguing about whether a given Leonardo was really a product of Leonardo's hand, for example.
This quick-and-dirty prompt returns results in line with my own understanding (which of course doesn't necessarily mean the exact figures are correct): https://chatgpt.com/share/698ce93b-b66c-800b-9e41-91ccab8eba... The artist I was primarily thinking of when I wrote that comment was Chihuly, whose workshop is famously more like a factory than a studio.
You know they cause a majority of the code of the core logic to execute, right? Are you sure the tests actually check that those bits of logic are doing the right thing? I've had Claude et al. write me plenty of tests that exercise things and then explicitly swallow errors and pass.
Yes, the first hour or so was spent fidgeting with test creation. It started out doing it's usual whacky behavior like checking the existence of a method and calling that a "pass", creating a mock object that mocked the return result of the logic it was supposed to be testing, and (my favorite) copying the logic out of the code and putting it directly into the test. Lots of course correction, but once I had one well written test that I had fully proofed myself I just provided it that test as an example and it did a pretty good job following those patterns for the remainder.
I still sniffed out all the output for LLM whackiness though. Using a code coverage tool also helps a lot.
" “Foreign persons” is broadly defined in the bill as any person who is not a United States person. However, the bill does not include “any corporation or partnership organized under the laws of a possession of the United States” in that definition of foreign person. "
Most offshoring companies don't pay directly to individuals - either they are employees of their own subsidiaries or those of offshoring entities. So, this is not going to change anything.
Are you sure of that? For years now I think Amazon and Apple have an agreement where only Apple or Apple-approved third-party vendors can sell Apple products on Amazon?
The replies to this thread cannot be serious, on a web forum populated—I thought—primarily by technologists. Surely you all remember the variations on, "If you make encryption illegal then only criminals will have encryption"?
The next step will surely be to make use of communication programs that law enforcement cannot read illegal, right? The police find some person who has committed a crime, caught in the ways that criminals are usually caught, such as with forensics, or simply with the guns and drugs in the boot of their car. Then they can see what forms of communication this person was using, and who was using it with them. At that point, it doesn't matter what those other people were doing: The use of banned encryption technology is the crime. You can roll them up for that, or use evidence of this crime to justify further intrusion into their meatspace lives. And so it goes, on up the chain of a criminal organization. Theoretically, at least.
I don't like this, I don't support this, but as has been said elsewhere in this thread: Let's not pretend this is some insurmountable problem for a government who has already shown an appetite for surveillance.
Sure, you could make unauthorized, fully encrypted communication illegal. But what would be the punishment for using it? Worse than for smuggling, human trafficking, murder? I seriously doubt it. If you're a criminal risking decades in prison for major crimes, using some illegal software is 100% worth it, if it significantly reduces the risk of getting caught for the real crimes you're committing.
You can't make laws that govern how criminals behave. All chat control will really accomplish is maybe a momentary string of arrests(which is meaningless in the long term; there's always someone to take over), and longer term, worse privacy and security for everyone except the criminals.
UK has the idea of contempt of court. Even as it stands, the court can demand you submit some evidence - say an encryption key for a document. And if you refuse, they can even imprison you until you surrender the key.
Another principle is that when someone is destroying evidence, you can presume it contained incriminating evidence.
I think you could make the punishment proportional to the presumed crime.
I was under the impression that, at least for models without "reasoning", asking them to be terse hampered their ability to give complete and correct answers? Not so?
> asking them to be terse hampered their ability to give complete and correct answers?
You can kind of guide both the reasoning and "final" answer individually in the system prompts, so you can ask it to revalidate everything during reasoning, explore all potential options and so on, but then steer the final answer to be brief and concise. Of course, depends a lot on the model, some respond to it worse/better than others.
