Navigating Change: ChatGPT's Web Browsing Feature Removal

7 Mar 2024 · 9 min

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

```markdown

AI Today Podcast Episode Notes

Navigating Change: ChatGPT's Web Browsing Feature Removal

Episode Overview In this episode of "AI Today," the hosts discuss the recent removal of the web browsing feature from ChatGPT, exploring the implications for users and the platform. They analyze the reasons behind this decision, the controversy it sparked, and how it affects user experience and the broader AI landscape.

Key Discussions

Background of the Controversy

  • Problematic Usage: Users were reportedly using ChatGPT to bypass paywalls by inputting links to articles from outlets like The New York Times. ChatGPT would then provide the full text of these articles.
  • OpenAI's Response: In light of this, OpenAI decided to disable the "browse with Bing" feature to prevent potential copyright violations and protect content owners.

Details of the Feature Removal

  • Announcement from ChatGPT:
  • The browsing capability was disabled due to concerns that it could inadvertently fulfill requests for full texts from paywalled sites.
  • OpenAI stated that they are working to fix the issue and hope to reinstate the feature soon.
  • User Impact: This change has led to inconvenience for users who relied on this feature for accessing updated information.

Analysis of the Situation

  • Implications of the Removal:
  • The browsing capability was considered one of the most valuable features of GPT-4, and its removal could affect user retention and satisfaction.
  • The hosts express concern over the concentration of power in large AI companies and the potential negative effects of abrupt feature removals on user workflows.

Discussion on AI Ethics and Content Ownership

  • Ethical Considerations:
  • The podcast touches upon the broader ethical implications of AI training on paywalled content.
  • An article from *The Economist* is referenced, highlighting that AI models have been trained on content from paywalled sources, raising questions about fair use and compensation for content creators.
  • Community Reactions:
  • Various reactions from users and industry professionals were discussed, with some arguing that an open internet benefits all while others maintain that paywalled content should remain behind payment.

Future Considerations

  • Anticipation of Feature Restoration:
  • The hosts express hope that OpenAI will restore the browsing feature soon, emphasizing its importance for paying users of ChatGPT.
  • Potential Changes in AI Functionality:
  • The conversation hints at the potential need for better regulations and practices regarding how AI interacts with content behind paywalls, as well as the need for diverse models in the AI landscape.

Key Takeaways

  • The removal of ChatGPT's browsing feature reflects ongoing tensions between AI capabilities and content ownership rights.
  • The podcast highlights the importance of user feedback and the need for AI companies to maintain transparency and responsiveness to community concerns.
  • The situation exemplifies the challenges faced by tech companies in balancing innovation with ethical considerations.

Additional Resources

  • Invest in AI Box: [AI Box Investment](https://republic.com/ai-box)
  • Join the AI Box Waitlist: [AI Box Waitlist](https://AIBox.ai/)
  • AI Facebook Community: [Join Here](https://www.facebook.com/groups/739308654562189)
  • Learn about AI in Music: [Musical AI](https://musicalai.pro/)
  • Learn more about AI Models: [AI Models Pro](https://aimodelspro.com/)

Closing Thoughts The episode emphasizes the need for a balanced approach as AI technologies evolve, encouraging listeners to stay engaged with ongoing discussions about AI ethics, content ownership, and the future of technology. ```

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00chat gpt just got caught up in a big controversy so today on the podcast we're going to talk about what it is they've actually made some what I think significant changes to ChadGPT, and I'm hoping that they're going to revert many of these, and hopefully some of these won't last for too long. But essentially what happened was a number of, I believe this was a week or two weeks ago, I saw a bunch of articles and a bunch of posts on Reddit and other places where people were claiming essentially they could go to companies like the New York Times or other journalists that had paywalled content. They could post the link to that paywalled content into chat chipt and say you know give me all the content on this page so if you ever hit a paywall you just use the link and chat chipt could go and grab you the information now there's a couple different interesting um ideas around this one being the fact that open ai probably paid for some sort of mass subscription potentially to access all the paywalled content on all the different journals that wouldn't be uh very far off um of course there's some people that say oh maybe open ai just paid for like one premium account on all the different services and was using that it's great but i find that highly unlikely i'm sure they have some sort of api access to some of those bigger ones in order to get around uh paywalls and whatnot but in any case a lot of people were using it for that and so in response to that chat gpt has just made a massive announcement and they've completely disabled the internet browsing the bing internet browsing plugin from chat gpt you can no longer use chat gpt to crawl the internet which was literally uh the best part about gpt4 one of the best features it had period that i always had enabled um it'd be useful for a lot of different things um and of course google bard actively does that as well so i think it's going to be really interesting to see what they do but right now if you go to chat gpt um and try to get it to do something it has a pop-up on chat gpt if you're in gpt4 that says we've temporarily disabled the browse with bing beta feature they shoot you over to they link you over to one of their blog posts which is just a couple sentences long but essentially it says chat gpt browse with bing is a beta feature available to chat gpt plus subscribers that allows chat gpt to search the internet to help answer questions that benefit from recent information we have learned that the gpt browser beta can occasionally display content in ways we don't want for example if a user specifically asks for a url's full text it might inadvertently fulfill this request as of july 3rd 2023 we've disabled the browse with bing beta feature out of an abundance of caution while we fix this in order to do right by content owners we are working to bring beta back as quickly as possible and appreciate your understanding now my personal take on all of this is i think this is a big overstep that they have to disable the entire plugin i'm assuming that the ability that bing has to scrape content from paywalled accounts ah you know okay so i guess here's the one caveat to this bing may have a special license essentially whereas a um whereas a web browser like google bing is able to go and scrape all the content on the paywalled accounts or the paywalled pages and so then it's essentially just using its like web crawler capabilities to display the content for free.

3:17If that's the case, I would say this is kind of shady. Well, I guess not shady. If that's the case, well, number one, it's shady, but number two, it shouldn't be, it might be actually harder to undo this. But the other thing is if they're just using some sort of API to, you know, like the Wall Street Journal, where they have some sort of mega enterprise account that allows them to display this kind of information, then they should just turn that off and not display any paywall information and the problem will be solved. What I suspect though is that it might be special access they have because Bing is a web browser specifically.

3:53So this isn't like if, you know, OpenAI created their own plugin to access all the content on the internet, essentially their own web browser, their web crawler. That I think would be a little bit easier because they would just have it set to not you know scrape anything paywalled in any case uh this is a big um deal for a lot of people it is very inconvenient when they pull access to something like this and i think that this is just another reason why it is so important with ai to not have these giant um monolithic um companies that essentially have all of the access and all the power in ai chai chimpiti is by and large the one-stop shop for uh text to ai um of course we have google bard that's coming up and i think that is very healthy i would love to see hundreds and thousands of different ai models in this space playing a big role because when one pulls a very critical feature like this browse feature which people may have been building into different tools um it really can throw a wrench into people's product or their workflows and what they're trying to do so it is fairly inconvenient it now i will say um something interesting you know they've obviously tweeted this out as well and a lot of people have responded to this tweet which is interesting someone jeff kirda keys said um inadvertently fulfill this request seems like a feature not a bug other llms are going to provide this regardless an open internet is a net positive paywalled sites will fall there's other places to get the same information for free i mean that's your opinion but also at the end of the day a paywalled site is a paywalled site and if they have a paywall you have to pay for or go somewhere else and i know people are like it shouldn't be free let's get around it but like they paid someone to create that content so i'm not gonna argue with that it it's paywalled is paywalled so it should should cost money if you know jeff made his website that had a paywall on it i'm sure he would like people to pay for his site if that's what he's asking them to do in any case um and it's not like a lot of these paywalled sites you know when they're doing like news stories it's not like you can't just go and find five other places.

5:58Now, something that's interesting is I actually tried this hack a while ago when I found out about it. And I wasn't actually able to be successful in that post in the Lincoln. It actually, it was a New York Times article that was paywalled and it said, sorry, there's an error. It wouldn't let me essentially view the article like I was trying to test it out and see if it would work. And so what I did instead was I took the title of the article, posted it in and said, go find this article online and give me the entire transcript of the article or whatever and it was able to do it when I said that so there's a couple workarounds um which of course there's no such thing as a workaround when the entire capability of searching on the internet with chat gpt is taken down so I'm very curious to see what moves they make in this regard and the other thing is it's not like like they just outright banned the entire feature today but what's interesting is this isn't like a new feature people have been talking about this for a little while now um it's kind of been this like hack so I'm not sure if chat gpt if it got to a boiling point where they felt like they had to take an action um if i'm not sure what the tipping point was if they just found out about it and decided to do it which would be kind of funny if you know a lot of if they don't have their ears and all the different reddit forums and different places where people are trying new features i imagine with 10 billion dollars they really should have a team that is kind of all over the internet looking at what's trending on twitter and reddit you know weeks and weeks ago and bring that to their attention but in any case i really hope they bring this feature back very soon because it was a major reason why a lot of people are paying for gpt4 and g chat gpt premium in any case i did want to bring up the fact that the economist in a paywalled article ironically wrote an article called artificial intelligence is reaching behind newspapers paywalls and essentially in the article they're saying that you know publishers have long accused tech firms of profiting from their content and now they have a point where they're saying you know essentially a lot of ai was trained off of paywalled content, right?

7:52Like right now, they're shutting off the Bing thing and saying like, okay, we're not going to show people paywalled content. But if you think about it, Chattributo is probably trained because they used Bing's web scraper. It was probably trained off of all of this now paywalled content for the New York Times and every other journalist. So this article actually came out back in March. So this is not a new issue, right? People have been kind of talking about the controversy around this problem for a while. And it seems like now it has kind of come to a boiling point so it's going to be really interesting to see how they move forward with this if they try to pull out paywall content from their ai model which i believe would be incredibly hard nigh impossible and what they try to actually do with the current situation with the uh the plugin that is now completely banned so really hoping that comes back and we will follow up and uh see when this new feature comes back and how they actually plan on addressing this situation in the future.

From the publisher

In this episode, we discuss how users and the platform are navigating the changes brought about by ChatGPT's decision to remove a popular web browsing feature, exploring the implications for user experience and platform dynamics.

See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

More from AI Today

All 897 episodes
Navigating Change: ChatGPT's Web Browsing Feature RemovalAI Today · 9 min
Listen in VO