flyval

Did you ask ChatGPT that?

Me: why did the commenter use the term “loop-closing”?

ChatGPT: The commenter used the term "loop-closing" to emphasize the concept of a feedback mechanism that corrects or validates information. In systems theory and engineering, a "closed loop" system is one where the output is continually fed back into the system as input for self-correction and optimization. This term is metaphorically applied here to describe a process where information is constantly reviewed and corrected, forming a self-regulating cycle.

In the context of Wikipedia, this loop-closing happens through community edits and oversight, where incorrect information is often quickly corrected by other users. The commenter is highlighting the absence of a similar, effective mechanism in Large Language Models (LLMs) to automatically correct or verify the information they produce. This comparison underscores a concern about the potential for LLMs to disseminate incorrect or misleading information without an inherent system for checking and correcting it. The choice of "loop-closing" as a term is to draw attention to this systematic difference in information validation between platforms like Wikipedia and LLMs.

We use language like “hallucinations, bullshit, etc.”, because it’s more accurate descriptors for what’s happening.

When I tell you the LLM hallucinated something, you know exactly what I mean. Same for “It’s rationalizing”, “It’s refusing to answer”, “It’s being illogical”.

Your desire to reduce it all down to stats is as a bad as someone arguing that we shouldn’t talk about balls and goals in basketball, because it’s all just atoms.

Huh? Can you explain the difference?

That’s like saying “Oh, you didn’t get the right answer to that math question, you just mimicked giving me the right answer”. Yeah, but if I gave you the right answer, then I gave you the right answer, and I must’ve gotten it somehow.

This is dumb. An individual human takes in more data than modern LLMs do.

https://open.substack.com/pub/echoesofid/p/why-llms-struggle...

I mean, google has access to ~all of that stuff anyway. Even if you’re self-hosting your email+calendar, everyone else isn’t.

I’d love to have more privacy on everything, but realistically, the ship’s sailed on most of it.

Have you actually tried it yourself? I’d recommend it. And I don’t mean just playing with it —- Try using it to help you build something. It’s much more efficient than googling and combing through stackoverflow. Hallucinations are not as common as you’re thinking.

You clearly can’t just take the code, paste it in, and trust that it works, but you shouldn’t be doing that with stackoverflow either.

> I don’t think it’s actually reasoning about things at all

This is a huge leap. There’s plenty of humans who couldn’t do that, especially historically.

Stop thinking about reasoning as a binary question and think of it as a spectrum. Actually, not even a spectrum, but a huge multi-dimensional problem. ChatGPT does better at some reasoning problems than most humans do, and worse at some others. You clearly found one that it’s particularly bad at.

> He doesn't say valuable, he says it matters more.

What does that mean? How are these words different in this context?

> Didn't the pandemic show that those jobs that matter aren't well paid?

What the… how in the world did you derive this from the pandemic? Seems like the phrase “essential worker” confused you. The people who were allowed to keep working in person were those that were essential in the short term to provide everyday services, that’s it. This says nothing about whether software developers matter more or less, since 99% of software developers kept working anyway.

But anyway, this argument is moving the goal posts — the original article was saying that you’ll get paid more for them.

> BTW The median salary in the NBA in 2021/22 is $4,347,600, so NBA > Tech

I think this is basically true when comparing individuals. Any one of the 450 players in the nba bring more enjoyment and meaning to the average persons life than the median software developer does. (Though obviously if you compare the entire tech industry to the nba, then tech will win out.)

This is trivially untrue:

- median hr salary: $61k

- median software developer salary: $111k

Maybe what you’re trying to say is that the average junior dev doesn’t spend enough effort on people. But that’s a very different statement than saying that understanding people is more valuable.

I think you’re really confused about inflation. The central bank doesn’t control inflation. Saying your currency is inflation proof doesn’t make sense, because you have no control over that.

Inflation is caused by the market valuing the currency less than it had before. For example, if people don’t trust you, your currency will undergo inflation.

Also, I see a lot of claims about metric dc being linked to MEUs but no explanation of what maintains the peg. You can’t just claim the peg, you have to enforce it somehow. Even the bank of London couldn’t do that in adverse conditions (https://www.investopedia.com/ask/answers/08/george-soros-ban...)