Stories
Slash Boxes
Comments

SoylentNews is people

posted by hubie on Wednesday December 14 2022, @06:34AM   Printer-friendly
from the we-need-AI-bots-to-check-for-AI-submissions dept.

The Q&A site has been flooded with ChatGPT coding answers that look correct but often aren't, with moderators calling for a halt:

Stack Overflow, a site where developers can ask and answer coding questions, has temporarily banned the use of text generated from ChatGPT, a chatbot released by Open AI last week.

[...] Since launching, it's been prompted in numerous ways, including to write new code and fix coding errors, while the chatbot can ask for more context when a human asks it to resolve coding problems, as OpenAI sets out in examples. But Open AI also notes that ChatGPT sometimes writes "plausible-sounding but incorrect or nonsensical answers."

This appears to be a key cause of its impact on Stack Overflow and its users who are seeking correct answers to coding problems. Additionally, because ChatGPT generates answers so quickly, some users are supplying lots of answers generated by it without parsing them for correctness.

[...] Stack Overflow says that ChatGPT answers have "swamped" its volunteer-based quality curation infrastructure because there are so many poor quality answers pouring in.

So far, Stack Overflow has detected posts generated by ChatGPT in the "thousands". The other problem is that many answers require a detailed analysis by someone with experience in the subject to determine if the answer is bad.

Previously:


Original Submission

 
This discussion was created by hubie (1068) for logged-in users only, but now has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 3, Interesting) by krishnoid on Wednesday December 14 2022, @07:06AM

    by krishnoid (1156) on Wednesday December 14 2022, @07:06AM (#1282346)

    "plausible-sounding but incorrect or nonsensical answers."

    Remember when they had that photo where people thought Melania was using a body double? I floated the "idea" that she never left New York, and you only ever saw her robotic doubles. She was trying to take over the world like Dr. Doom, and when you saw Melania next to Donald, she was the server rack and he was her AI text-generation-to-speech project in alpha-test.

    What does that have to do with taking over the world? Nothing, really -- but most people need at least *one* hobby.

  • (Score: 0) by Anonymous Coward on Wednesday December 14 2022, @08:47AM (2 children)

    by Anonymous Coward on Wednesday December 14 2022, @08:47AM (#1282352)

    "plausible-sounding but incorrect or nonsensical answers."

    Sounds like many Stack Overflow answers.

    But yeah this may increase the amount of incorrect answers from >50% to >90%?

    Seriously, the first bunch of AI generate code is probably going to be very similar to this. A high chance of plausible but wrong or exploitable stuff.

    • (Score: 4, Touché) by acid andy on Wednesday December 14 2022, @11:56AM (1 child)

      by acid andy (1683) on Wednesday December 14 2022, @11:56AM (#1282362) Homepage Journal

      Seriously, the first bunch of AI generate code is probably going to be very similar to this. A high chance of plausible but wrong or exploitable stuff.

      That's what always happens when you try and replace a skilled, talented programmer with the cheapest possible alternative. It's just this time it's an AI. Still, most corporations no longer seem to care that their products these days are broken trash that is completely unfit for purpose, so why should software be any different?

      --
      "rancid randy has a dialogue with herself[...] Somebody help him!" -- Anonymous Coward.
      • (Score: 2) by krishnoid on Wednesday December 14 2022, @09:01PM

        by krishnoid (1156) on Wednesday December 14 2022, @09:01PM (#1282423)

        Because software is easily modifiable once bugs are found -- it doesn't require physical retooling, remanufacturing, and reshipping. At least that's what I used to think, until realizing that software being as buggy as it is even with those insurmountable advantages, is still one of my go-tos when I start to think that I might understand how product marketing works.

        SQLite [sqlite.org] is an example of good software, so you know it's possible, but like Hondas/Toyotas, the LaserJet 4250 [printertechs.com], they tend to be isolated cases rather than industry standards.

  • (Score: 1, Flamebait) by Frosty Piss on Wednesday December 14 2022, @12:02PM

    by Frosty Piss (4971) on Wednesday December 14 2022, @12:02PM (#1282363)

    plausible-sounding but incorrect or nonsensical answers

    In other words, not much has changed. It's not a totally useless resource, but beware. And it's a bit like Wikipedia where you have these little princes defending their territory from their mom's basement.

  • (Score: 2) by bzipitidoo on Thursday December 15 2022, @02:18AM

    by bzipitidoo (4388) on Thursday December 15 2022, @02:18AM (#1282451) Journal

    Don't know how bad these problems are on Stack Overflow, but on Quora, I've encountered insincere questions. The person asking isn't exactly interested in a plain answer. In one case, I suspected it to be a shill trying to promote MS products. Asked how to convert data to some format I'd never heard of before, and when I searched it, found out that it was an MS format, supposedly open. The askers could have searched that themselves. Seemed likely they were asking only to boost that format's popularity. Wouldn't surprise me if they had a bot spamming Quora participants with variations on that question. In another case, the question sure sounded like a homework assignment. In still another case, the question was how to construct a small zip file that when expanded would be so huge it would greatly exceed the capacity of the average data storage device. In other words, they wanted what is known as a zip bomb. Why would anyone want that? Not likely for research purposes!

(1)