Suggestion Can ChatGPT discussions be subdivided into categories?

  • Thread starter Thread starter Stephen Tashi
  • Start date Start date
  • Tags Tags
    chatgpt
AI Thread Summary
The discussion centers on the need to categorize conversations about ChatGPT and other AI bots due to the increasing volume of threads. Participants express differing opinions on the usefulness of ChatGPT, with some valuing its ability to answer straightforward questions while others question its reliability and the quality of discussions surrounding it. There is a call for separating discussions that critique AI outputs from those that focus on their practical applications. Concerns are raised about the appropriateness of citing ChatGPT as an authoritative source, with some arguing that it should not be compared to reliable tools like chess programs. Ultimately, the conversation highlights the diverse perspectives on the role and reliability of AI in technical discussions.
Stephen Tashi
Science Advisor
Homework Helper
Education Advisor
Messages
7,864
Reaction score
1,602
Having used ChatGPT several times, I predict it will become an extremely popular topic of conversation. Threads about such bots are getting too long, so I suggest subdividing discussions of ChatGPT and other online AI bots into categories.

On the one hand, some people enjoy posting about puzzles that expose the failings of AI bots. On the other hand, I find ChatGPT useful for answering straightforward questions that can, in theory, be answered by doing a lot of online browsing and cross referencing. (e.g. Which species of juniper tolerate clay soil? Which of those species of Juniper grow to tall tress? Are any of those species dioecioius? Do white winged doves eat the berries of those species?) Discussions about tricking AI bots should be divided from discussions about the quality of their (at face value) useful replies.

There has been discussion a the lines of "ChatGPT is not an authoritative source" etc. I suppose there are sections of the forum where people frequently cite authoritative sources. In the sections I frequent, I haven't noticed this. At any rate, quoting a ChatGPT chat for purposes of critique isn't the same a citing it as an authoritative source. So I hope a ban on ChatGPT as a source doesn't imply a ban of quoting ChatGPT as a topic for discussion.
 
Physics news on Phys.org
Stephen Tashi said:
At any rate, quoting a ChatGPT chat for purposes of critique isn't the same a citing it as an authoritative source. So I hope a ban on ChatGPT as a source doesn't imply a ban of quoting ChatGPT as a topic for discussion.
The rules allow “AI-generated text output in entertainment threads in General Discussion” and “threads about the technology and cultural impact behind AI like ChatGPT”. Do these cover what you’re thinking about?
 
Nugatory said:
Do these cover what you’re thinking about?
No. Those topics avoid serious discussion of the content of ChatGPT generated material. (I take "entertainment" as diverging from something serious.)

For example, in a math section, I can quote material from somebody's blog post and ask technical questions about it. In doing that, I'm not necessarily quoting an auhoritative source.
 
Stephen Tashi said:
For example, in a math section, I can quote material from somebody's blog post and ask technical questions about it.
As long as the quoted material is not too crackpotty, yes. We’d be discussing some thoughts/ideas relevant in that section, and that’s a reasonable thread start.
But text created by ChatGPT doesn’t represent any thoughts/ideas, it’s just words mechanically arranged to resemble other text somewhere on the internet. Asking technical questions about it makes about as much sense as asking technical questions about the output of an Ouija board.
 
  • Like
Likes Vanadium 50, Bystander and weirdoguy
Nugatory said:
Asking technical questions about it makes about as much sense as asking technical questions about the output of an Ouija board.
I disagree (about the output of an AI bot being comparable to the output of a Ouija board).

That's analogous to saying the output of a chess playing program is not worthy of quoting - on the grounds that computer programs don't really think about chess.
 
  • Skeptical
  • Like
Likes weirdoguy and PeroK
Why does everything need to be discussed in PF? One would assume that there are higher quality discussions of this elsewhere. I know that I have not been impressed in general with the local ChatGPT level of discourse.
 
Frabjous said:
I have not been impressed in general with the local ChatGPT level of discourse.

Are talking about discussions focused on ChatGPT itself?

I won't comment on the quality of those discussions because I haven't read many of them. My post is about separating that type of discussion from discussions of other technical topics.
 
ChatGPT-4 has been a game changer for me with technical skills. I've been able to fix and optimize my SQL queries and Python code faster than ever. No more stack overflow.

One process I wrote in Python was extremely slow to the point of being unusable. I asked ChatGPT-4 to convert it to a SQL query. It worked and now functions quickly.
 
  • Like
Likes dextercioby
Greg Bernhardt said:
ChatGPT-4 has been a game changer for me with technical skills. I've been able to fix and optimize my SQL queries and Python code faster than ever. No more stack overflow.

One process I wrote in Python was extremely slow to the point of being unusable. I asked ChatGPT-4 to convert it to a SQL query. It worked and now functions quickly.
Better than asking a ouija board, dishwasher or rock to fix your code?
 
  • #10
Stephen Tashi said:
That's analogous to saying the output of a chess playing program is not worthy of quoting - on the grounds that computer programs don't really think about chess.

But chess playing programs were designed to be reliable when it comes to chess. ChatGPT was not designed to be reliable in physics/maths.
 
  • #11
The observation I keep making about ChatGPT is, to steal a line from Morris Udall, "everything that needs to be said has been said. But not everyone has said it yet."
 
  • Like
Likes russ_watters, BillTre, Nugatory and 2 others
Back
Top