With the advent of new artificial intelligence (AI) tools such as ChatGPT, we are experiencing a colossal technological transformation at the level of information, radically changing the conditions for our access to knowledge. Until recently, search engines have been one of the main sources of knowledge for Internet users. We have all turned to them to find the sources we need to understand the world. Now with ChatGPT the situation is changing dramatically, since this tool can completely replace the search engine.

So until yesterday, visiting her search engine Google in essence, we were given a list of hyperlinks sorted by popularity, and we chose which ones to follow. In addition to the criterion of popularity, the search engine had and still has two more arbitrary elements that deliberately distort our information. First, Google initially displays hyperlinks that are the product of a commercial transaction. As a rule, the first results of any search are ads. Secondly, there is a clear bias in favor of the first page of results. It has been measured that very few people go to the second or third page of results – the vast majority stop at the first search results. These three parameters (popularity, advertising, first page) prevent the search engine from becoming the deepest and most unbiased repository of human knowledge and, therefore, a completely reliable source of information.

Timely action must be taken by the political system, which, unfortunately, tends to rank such matters low.

A new tool for knowledge generation that we met a few months ago is based on artificial intelligence, it is called ChatGPT and impresses with its efficiency. Just a few days ago, its 4th edition was unveiled, which uses machine learning technology to mimic the synapses of the human brain. This neural network receives a lot of data, it is able to detect patterns and predict the next matching word in text or the next matching pixel in an image. Advances in this form of AI, called “generative”, are associated with the emergence of a new generation of algorithms – “transformers”. The goal of this algorithmic process is to create original texts or images based on simple natural language commands. The relevance for the work of search engines is obvious. Both tools ask us a simple question and then in turn give us a good answer. The ChatGPT text generator does not provide hyperlink answers. He composes regular texts that have a beginning, middle and end, removing the “noise” of multiple or even conflicting sources. ChatGPT combines everything into a single text, freeing the user from painstaking and creative drafting work. In fact, he offers him a special composition on a plate. The first very serious problem is that in ChatGPT we don’t see what the sources of the text are. At Google, we knew that when we clicked on a link, we were reading a person’s blog or a New York Times page. This very important piece of information is left out here. The user not only cannot know the source, but also cannot distinguish whether the result is advertising or not. ChatGPT synthetic text generation is the definition of a “black box”.

This area has been bypassed by the generative AI tool Perplexity, which displays the source after each sentence of its composition, which is really very useful. The same thing happens in the still experimental version of the Microsoft Bing search engine: hyperlinks are also visible there. Of course, the problem has not been solved here either, because the publication of sources does not automatically mean reading them. I predict even fewer people will click on sources.

ChatGPT takes a step back in terms of visibility and breadth of information on the web. Because of its opacity, but also its dubious effectiveness, it will further alienate users from multiple sources and research, dramatically increase the amount of misinformation, and further cloud the information landscape. Perhaps some technical solutions will help, but at the same time, users need to be trained. The fact that search engines are experimenting with this tool in order to seamlessly integrate it into their main product cannot but cause concern. Before taking the next step, there must be time for consultation and key decisions. We have a bad example of social media that has gradually become a nightmare for modern democracies. In order not to once again test the limits of social endurance, the political system must take timely measures, which, unfortunately, usually underestimate such issues.