Hacker News new | ask | show | jobs
by setgree 1207 days ago
As is often true of GPT responses, there's some nonsense interspersed here, e.g. the claim that R has "a more mature package universe" than Python.

I think this is false, but if you're reading quickly, it sounds cogent enough. As Sarah Constantin observed about GPT2 [0]:

> if you skim text, you miss obvious absurdities. The point is OpenAI HAS achieved the ability to pass the Turing test against humans on autopilot...The mental motion of “I didn’t really parse that paragraph, but sure, whatever, I’ll take the author’s word for it” is, in my introspective experience, absolutely identical to “I didn’t really parse that paragraph because it was bot-generated and didn’t make any sense so I couldn’t possibly have parsed it”, except that in the first case, I assume that the error lies with me rather than the text. This is not a safe assumption in a post-GPT2 world. Instead of “default to humility” (assume that when you don’t understand a passage, the passage is true and you’re just missing something) the ideal mental action in a world full of bots is “default to null” (if you don’t understand a passage, assume you’re in the same epistemic state as if you’d never read it at all.)

[0] https://www.skynettoday.com/editorials/humans-not-concentrat...

4 comments

> there's some nonsense interspersed here, e.g. the claim that R has "a more mature package universe" than Python.

As a programmer, I find R hard to use and not very well designed, so I can see why you'd call that nonsense.

But when I was a math student, I found that in some ways R does have "a more mature package universe". There were many math algorithms that I could find packages for in R and not in Python, even as a mere grad student.

Absolutely, for statistics and visualization I think R and its packages are (sometimes) superior. But GPT responses don't generally offer those kinds of nuances; the claim is that the packages are "more mature," period. And it's for good reason that the _most_ mature Python packages, e.g. numpy and pandas, are used by data scientists in production pretty much everywhere.
amazingly, your comment will eventually be added to the CHatGPT corpus and at some point down the line may be used to add the nuance that's currently lacking :)
Assuming it's not a GPT response
Regarding numpy/pandas: What's the reason outside of them being _in Python_?
I wonder if the "default to humility" heuristic probably does more harm than good on net, because the people who heed it probably shouldn't, and the ones who should won't.
Default to humility. Do not assume you're so smart that you can skim the text and understand it correctly. Read every word, don't assume that the author is so predictable that you can guess correctly.

Why, does not sound too arrogant to me.

I think it's important to remember that Humans who are not-too-smart can also sound coherent, yet also babble complete nonsense.

My experience with ChatGPT thus far is that it is as intelligent as a very broadly read person who just doesn't reeeally get the complex or nuanced aspects of the content it reads - much like many real Humans.

Robin Hanson makes this point in better babblers http://www.overcomingbias.com/2017/03/better-babblers.html

"After eighteen years of being a professor, I’ve graded many student essays. And while I usually try to teach a deep structure of concepts, what the median student actually learns seems to mostly be a set of low order correlations. They know what words to use, which words tend to go together, which combinations tend to have positive associations, and so on. But if you ask an exam question where the deep structure answer differs from answer you’d guess looking at low order correlations, most students usually give the wrong answer."

Reminds me how when people get criticized on Twitter now, they just assume it’s a bot