ChatGPT responses are banned from StackOverFlow...for now

Amongst all the hype around ChatGPT, Stack Overflow has made a decision to temporarily ban users from sharing responses generated by the AI chatbot.

a year ago   •   4 min read

By Peter Foy

Amongst all the hype around ChatGPT, Stack Overflow has made a decision to temporarily ban users from sharing responses generated by the AI chatbot.

The site's moderators stated that the ban is temporary and that a final ruling will be made in the future after consultation with the community.

One of the main reasons for this ban is the ChatGPT makes it so incredibly easy for users to generate responses to coding questions and flood the site with answers that may look right at first glance, but are often wrong after a closer look.

As the mods write:

Overall, because the average rate of getting correct answers from ChatGPT is too low, the posting of answers created by ChatGPT is substantially harmful to the site and to users who are asking or looking for correct answers.

The mods continue to say that:

The primary problem is that while the answers which ChatGPT produces have a high rate of being incorrect, they typically look like they might be good and the answers are very easy to produce.

Stay up to date with AI

We're an independent group of machine learning engineers, quantitative analysts, and quantum computing enthusiasts. Subscribe to our newsletter and never miss our articles, latest news, etc.

Great! Check your inbox and click the link.
Sorry, something went wrong. Please try again.

What led to ChatGPT being banned on StackOverflow?

ChatGPT is based on OpenAI's autocomplete text generator GPT-3.5, and was released last week. It generates text and code by analyzing patterns from massive amounts of text scraped from the web (read: the entire internet).

While many have been using ChatGPT to generate interesting screenplays, funny jokes, and cute poems, StackOverFlow is a much different beast.

The moderators said the:

...volume of these answers (thousands) and the fact that the answers often require a detailed read by someone with at least some subject matter expertise to determine that the answer is actually bad has effectively swamped our volunteer-based quality curation infrastructure.

In short, the site's volunteer-based quality curation infrastructures requires a detailed analysis by someone with experience in the subject to determine if the answer is incorrect, and they simply don't have the capacity to read all the AI-generated submissions.

As a consequence, if a  user is caught posting answers generated by ChatGPT, Stack Overflow will impose sanctions on the user, even if the post would have been acceptable.

What is ChatGPT?

In case you haven't been obsessed with ChatGPT for the past week, here's a quick overview of the chatbot from our guide on what is ChatGPT:

  • ChatGPT is a prototype dialogue-based AI chatbot that has the ability to understand natural human language and generate detailed human-like written text.
  • Developed by OpenAI, ChatGPT has been described as an alternative to Google because it can provide answers to complex questions and also has the potential to be used for generating content for websites, answering customer inquiries, and much more.
  • ChatGPT has limitations such as a knowledge base only extending up until 2021, which means that certain queries and searches are simply impossible for the chatbot to answer.

Another key limitation that directly relates to StackOverflow's situation:

The AI has been known to produce incorrect or nonsensical responses, and may even present misinformation as factual.

How does ChatGPT work?

OpenAI trained its AI chatbot, ChatGPT, using reinforcement learning from human feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection setup.

  • ChatGPT was fine-tuned from a model in the GPT-3.5 series, which completed training in early 2022. It was trained using Azure AI supercomputing infrastructure.
  • A Large Language Model, or LLM, is a class of AI model that has been trained on an extensive dataset of text data in order to generate written text that resembles human language. ChatGPT is one such LLM.

Use Cases of ChatGPT

ChatGPT has a range of potential real-world use cases, such as generating website content, responding to customer inquiries, and more.

It is also suggested that professions reliant on content creation could be made obsolete by this technology, however ChatGPT's limitations currently limit its utility in fields such as journalism.

Responses from StackOvervflow users

Overall it seems liek many people are happy wiht this decision by StackOverflow, as the top comment writes:

Well done! Glad you made the right decision and really hope it will become permanent and be extended to ban any AI generated answers. AI will never be able to post good programming answers, not even in 100 years.

Definitely don't know about that last part...but we'll see :)

On the other hand, users are saying

It should not be temporary.

Where do we go from here?

The temporary ban of ChatGPT on Stack Overflow provides a good example of the current limitations of AI-generated code and text.

While the chatbot may be able to generate impressive text (albeit often quite bland), its tendency to produce incorrect or misleading responses has shown the potential to be harmful to users seeking accurate answers.

As the field of AI and more specifically generative AI continues to advance, it will be crucial for moderation systems to adapt in order to prevent the proliferation of erroneous and misleading content.

In the meantime, we can only hope that ChatGPT's creators will continue to fine-tune the AI's abilities in order to better serve the needs of users.

In regards to the question of whether AI will take over coding jobs, it looks like they're safe...for now.

Spread the word

Keep reading