0
0
OpenAI has unveiled a significant upgrade for ChatGPT. They’ve announced that they are extending the model’s functionality to include web browsing features exclusively for ChatGPT Plus and Enterprise subscribers.
This new (not so new actually they first launched the feature in May and optimized the feature based on user feedback) feature allows the chatbot to explore the web using Bing. So the model now can provide users with concise summaries of the results and source links.
It is exciting because ChatGPT is no longer confined to the training data predating September 2021. Now the tool can deliver the most current information on topics ranging from current events to news and research.
Pros and Cons of This Upgrade
Now the upgrade empowers users to access more up-to-date information. This allows OpenAI to better compete with rivals like Google Bard, known for their real-time web search capabilities. However, the accessibility of web browsing is currently limited to Plus and Enterprise users, with a $20 monthly subscription cost for Plus. This subscription cost stands in contrast to the free accessibility of competitors like Bard and Bing Chat.
While OpenAI does plan to extend web browsing to all users in the future, the timeline for this expansion remains unspecified. Also not so bright but, users are required to consent to share their browsing data with OpenAI. Some are concerned about privacy and data security.
Inconsistent Performance and Technical Hurdles
Regrettably, availability, and privacy are not the only concerns raised by this browsing feature. Since its launch, several users have reported issues with ChatGPT’s web search abilities. Some have even pointed out instances where the bot claimed not to know answers without attempting a search or failed to access certain web pages due to restrictions.
These issues are compounded by the fact that many popular websites have chosen to block OpenAI’s web crawler, GPTBot, citing copyright concerns. Currently, a quarter of the top 1,000 websites, including major publishers like The New York Times, The Guardian, CNN.com, USAToday, and others, have blocked GPTBot’s access to their content.
Copyright Concerns and Lack of Transparency
OpenAI’s decision to connect ChatGPT to the internet has sparked concerns about the content being scraped from the web. Questions have arisen about the inclusion of personal data, copyrighted materials, or intellectual property from third-party websites in ChatGPT’s training data.
One of the major challenges here is OpenAI’s black box approach to AI development, which lacks transparency regarding the data used to train ChatGPT. This lack of clarity makes it difficult to determine if protected data categories such as personally identifiable information (PII) or intellectual property (IP) have been incorporated into the training data.
The Final Verdict
The connection of ChatGPT to the internet brings promising new capabilities for Plus and Enterprise users. However, it’s not without its challenges, particularly concerning the substantial number of sites blocking GPTBot. As long as concerns regarding AI models processing copyrighted material and IP continue, these limitations are likely to endure.
This development is significant, but it comes with its set of difficulties and complexities.
