“Key Takeaways: OpenAI’s choice to connect ChatGPT to the Internet opens up some fascinating new use cases for Plus and Enterprise customers, but third-party website bans make it rough around the edges.”
Earlier this week, OpenAI announced the integration of web browsing capabilities for ChatGPT Plus and Enterprise users. With this new plugin, the chatbot can search Bing, provide users with a summary of the result, and share a link to the source.
As we can see with these type of advancements, the question arises: Does ChatGPT has the ability to destabilize white collar work?
“Browsing is now available to Plus and Enterprise subscribers, with plans to expand to all users in the near future. To activate, select Browse with Bing under GPT-4.
— OpenAI (@OpenAI), September 27, 2023”
This latest update means that ChatGPT is no longer restricted to training data from before September 2021. It can now provide users with the latest information on current events, news, and research.
To enable this feature, ChatGPT Plus and Enterprise users can go to Profile & Settings, select Beta features, and choose Browse with Bing under the GPT-4 option.
The announcement follows OpenAI’s previous introduction of web browsing as part of its ChatGPT Plus plan. However, this feature had to be withdrawn due to concerns about users bypassing paywalls.
“We discovered that ChatGPT’s “Browse” beta can display content in unexpected ways, such as when a user directly requests the whole text of a URL, which it may mistakenly fulfill. We’re disabling Browse while we address issue because we want to do the right thing by the content owners.
— OpenAI (@OpenAI), July 4, 2023”
It also comes just days after OpenAI added multimodal voice and image capability to ChatGPT.
The Good & Bad News
OpenAI’s decision to connect ChatGPT to the internet has both positive and negative aspects. On the positive side, it gives users access to more current information, which enhances the solution’s usefulness. This move also positions OpenAI to compete more effectively with rivals like Google Bard, which can search the internet for real-time responses.
However, limiting web browsing to Plus and Enterprise users restricts accessibility. The $20 per month cost for a Plus subscription, compared to the free availability of Bard and Bing Chat, seems steep.
While OpenAI plans to expand web browsing to all users in the future, it hasn’t provided a timeline for this change.
Users must also opt in to enable their chat history to use the plugin. This means they must agree to share their browsing data with OpenAI, raising new concerns about privacy and data security.
You can learn more about navigating restrictions & overcoming challenges of ChatGPT.
Inconsistent Performance
Inconsistencies in performance have become apparent with the browsing plugin. Several users have reported issues with ChatGPT’s ability to search the web since its release.
For example, Jeremy Howard, co-founder of FastDot AI, tweeted about his experience: “ChatGPT Web Browsing kept telling me it didn’t know stuff, without bothering to search at all! And then when it did search, it claimed to not find anything.”
“ChatGPT Web Browsing constantly told me it didn’t know things without even searching! Then, when it searched, it said it had found nothing.@bing tries to be a good Bing here (but makes up something that’s not real), and ChatGPT doesn’t even try to be good! 🙁
— Jeremy Howard (@jeremyphoward), September 28, 2023”
ChatGPT Error In Body Stream is common type of error that users face while using ChatGPT.
Other users have expressed similar frustrations. Sully Omar, co-founder and CEO of Cognosys, shared an example of ChatGPT failing to list the top 5 waterproof shoes “due to restrictions on accessing certain web pages.”
“A little dissatisfied with chatGPT browsing:
— Sully (@SullyOmarr), September 28, 2023”
While it’s challenging to pinpoint the exact cause of these performance issues, it’s crucial to recognize that ChatGPT’s search capabilities are limited. This limitation stems from the fact that many popular websites have blocked OpenAI’s web crawler, GPTBot, from accessing their content due to copyright concerns.
Currently, 25% of the top 1,000 websites have blocked GPTBot. These include major publishers like the New York Times, The Guardian, CNN.com, USA Today, Business Insider, Reuters, Washington Post, NPR, CBS, NBC, Bloomberg, CNBC, ESPN and USA Timely.
Moreover, while connecting ChatGPT to the internet expands its capabilities, it doesn’t eliminate the inherent risk that large language models (LLMs) like ChatGPT have of generating incorrect information and facts, a phenomenon known as hallucination.
Other Issues: Copyright Infringement
With ChatGPT now connected to the Internet, significant concerns have emerged regarding the content OpenAI is scraping. People are questioning whether this scraping includes personal data, copyrighted materials or intellectual property from third-party websites.
One major challenge is OpenAI’s opaque approach to AI development, which provides no transparency into the data ChatGPT is trained on. This lack of transparency means the public can’t know if protected data categories like personally identifiable information (PII) or intellectual property (IP) are part of the training data.
While websites can opt out of having their data scraped by GPTBot, they must take action to do so.
It’s important to note that this challenge isn’t unique to OpenAI. Other providers, such as Google, also allow websites to opt out of training their AI models.
The Bottom Line
Connecting ChatGPT to the Internet opens up new possibilities for Plus and Enterprise users, but there are still some issues to iron out.
The large number of sites blocking GPTBot severely limits ChatGPT’s web browsing capabilities. This situation is unlikely to improve soon as concerns about AI models processing copyrighted material and intellectual property persist.
Read more Articles Related to ChatGPT:
