ChatGPT gets access to the Internet – here’s what to watch out for

By
sachith
-
0
238

The well-known AI chatbot is allowed to surf. What are the consequences?

OpenAI, the American company behind ChatGPT, has announced that it will allow ChatGPT back online. The AI has so far only been fed data until 2021. What does this mean for users?

All new changes to ChatGPT at a glance

ChatGPT now browses: On Twitter, the developers wrote that ChatGPT now has access to the Internet and you can browse with ChatGPT’s help. In its replies, the AI chatbot now additionally provides links to sources.

However, there is a small catch. Access to this new feature called “Browse with bing” is currently limited to the paid upgrade of ChatGPT Plus and Enterprise

Since the developers are supported by Microsoft, the use of Bing is unsurprising. This is the search engine from the house of Gates. Microsoft’s AI bot has supported searching the web for some time now.

However, ChatGPT developers promise that this feature will soon be available to all users.

A few days ago, OpenAI also announced that it will soon be possible to interact with the AI via voice control.

What’s so special about it?

It was possible to ask ChatGPT about various topics even before that. However, his knowledge was limited to the period up to September 2021 and no links were provided when answers were given.

If you ask ChatGPT now for example about news of the last weeks, he doesn’t answer anymore with the message that he can’t provide real-time information.

This concept is not entirely new, though. Google Bard also shows links in response to queries and can play back the answer via voice output.

Nevertheless, it is important that this step is also taken with ChatGPT.

What is the danger of real-time data?

(The Internet is like the real world. It is a web of constantly changing information. It builds new links and gets new content all the time. (Symbol image; photo by NASA on Unsplash))
(The Internet is like the real world. It is a web of constantly changing information. It builds new links and gets new content all the time. (Symbol image; photo by NASA on Unsplash))

With a closed data set that can be accessed by an AI, it is much easier to filter out malformed responses from a source control perspective. Here, developers can look very carefully for bad content and then exclude it more consistently. The quality of the data set is therefore correspondingly high.

Accessing a permanently changing network with thousands of new contents per second therefore involves a certain risk. A query is therefore more likely to result in an incorrect or risky answer. The potential for misuse increases.

It is therefore important that sources are now provided in addition to the answers. The user can look at these and decide for himself, based on the original source, how valuable the information received is.

What do you think? Is it an important and right step to know exactly where the AI is actually searching in order to evaluate the source? What are the dangers of searching the living web? Do you think ChatGPT is easier to manipulate now? Or do you rather think that the current ChatGPT is too limited? Feel free to write us your opinion in the comments.